SCADA systems

Why it matters

Water supply networks, oil and gas pipelines, electricity transmission and distribution, metro rail power and irrigation canals are spread over tens to thousands of kilometres. Supervisory control and data acquisition (SCADA) lets a few operators in one control centre watch and command all of it. SCADA engineering — RTUs, communication links, protocols and cybersecurity — is a large employer of instrumentation graduates in India's smart-city, power and water sectors.

Key ideas

What SCADA does. It acquires data (analog values, statuses, counters, alarms) from remote sites, presents it to operators, stores it, and sends supervisory commands (start a pump, open a breaker, change a set point). Fast closed-loop control normally stays local, in the RTU or PLC at the site; SCADA supervises it.

Components.

• Field instruments and devices — transmitters, flowmeters, valve actuators, breakers, pump starters.
• RTU (remote terminal unit) or PLC at each site — reads I/O, runs local logic, time-stamps events, buffers data, and communicates with the master. RTUs are built for wide temperature ranges, low power (often solar) and weak links. In substations, IEDs (intelligent electronic devices such as protection relays) also act as data sources.
• Communication system — licensed radio, GSM/4G, fibre, leased lines, power-line carrier, VSAT satellite. Links may be slow, costly and intermittent.
• Master station — redundant SCADA servers with front-end processors, a real-time database, historian, alarm and event processing, and HMI clients in the control centre. In power systems it is extended into an energy management system (EMS) or distribution management system (DMS).

Data collection methods.

• Polling — the master asks each RTU in turn. Simple and deterministic; the polling cycle time grows with the number of RTUs and the data per RTU.
• Report by exception (unsolicited reporting) — the RTU sends only values that changed by more than a deadband, plus periodic integrity polls. Much less traffic and faster alarm delivery on slow links.
• Sequence of events (SOE) — status changes are time-stamped at the RTU to millisecond resolution (with GPS-synchronised clocks), so the order of breaker trips can be reconstructed regardless of when the data reaches the master.
• Store and forward — data is buffered in the RTU during a link outage and back-filled into the historian later.

Protocols. Modbus RTU (serial) and Modbus TCP — simple master/slave register reads; DNP3 — widely used in water and power, supports time-stamped events, unsolicited reporting and data classes; IEC 60870-5-101/104 — telecontrol for power grids; IEC 61850 — substation automation. Serial characters usually carry start, stop and parity bits, so a byte costs 10 or 11 bits on the line.

SCADA versus DCS. SCADA is event- and data-driven, works over wide areas and unreliable links, and leaves control local. A DCS is a tightly integrated, high-speed system inside one plant that executes the control loops itself. The distinction blurs as both use Ethernet and the same HMI software.

Cybersecurity. SCADA once relied on isolation ("air gap"); today's networks are connected, and attacks such as Stuxnet (2010) and the Ukrainian grid attacks showed the risk. Good practice follows IEC 62443: zones and conduits, firewalls and a DMZ between corporate IT and the control network, secure remote access, authentication on protocols (DNP3 Secure Authentication), patch and backup management, and monitoring.

Formulas

t_tx = N_bits / R

• t_tx: transmission time (s); N_bits: bits sent including framing (start, parity, stop bits); R: line bit rate (bit/s).

N_bits = N_bytes × b_char

• b_char: bits per character on the line (10 for 8-N-1, 11 for 8-E-1 or 8-N-2).

T_cycle = Σ (t_request + t_response + t_turnaround) over all RTUs

• Polling cycle time (s); t_turnaround: RTU response delay plus modem/radio key-up time.

R_required = n_values × bits_per_value × updates_per_second × (1 + overhead)

• Data rate needed for continuous updates (bit/s).

Worked examples

Example 1 (standard) — link loading. An RTU must send 200 analog values every second; each value is 16 bits, and protocol overhead adds 25 %. (a) What bit rate is needed? (b) Can a 9600 bit/s radio link carry this?

1. R_required = 200 × 16 × 1 × 1.25 = 4000 bit/s.
2. Link utilisation = 4000/9600 = 0.417 (about 42 %).
3. The link can carry it in principle, but with other RTUs sharing the channel, retries and turnaround delays, report-by-exception with deadbands would be the safer design.

Example 2 (GATE level) — polling cycle. A master polls 50 RTUs at 19 200 bit/s. Each poll (request + response) is 40 bytes in total, sent as 11-bit characters, and each RTU adds a turnaround delay of 30 ms. (a) Find the time per RTU and the full polling cycle. (b) What is the worst-case delay before a new alarm at an RTU reaches the master?

1. Bits per poll = 40 × 11 = 440 bits.
2. Transmission time per RTU = 440/19 200 = 0.0229 s = 22.9 ms.
3. Time per RTU = 22.9 + 30 = 52.9 ms.
4. Polling cycle = 50 × 52.9 ms = 2.65 s.
5. (b) An alarm that occurs just after its RTU has been polled waits almost a full cycle before the next poll, so the worst-case delay is about 2.6 s plus the poll itself (≈ 2.7 s). Unsolicited reporting (e.g. DNP3) would deliver it within one message time, and SOE time-stamps preserve the true event time either way.

Common mistakes

• Using 8 bits per byte for serial links and forgetting start, parity and stop bits.
• Putting fast control loops in the SCADA master instead of the RTU/PLC; a link failure then stops control.
• Polling everything at the same fast rate instead of grouping data by priority and using report by exception.
• Using the arrival time at the master as the event time; use RTU time-stamps for sequence of events.
• Treating the air gap as security; connections through laptops, USB drives and vendor remote access bypass it.
• Ignoring clock synchronisation between RTUs, which makes SOE logs useless.

For GATE IN

SCADA appears mainly in conceptual questions: components and their roles, polling versus report by exception, SCADA versus DCS, common protocols and communication media. Numericals are short data-communication calculations — transmission time, bit rate, polling cycle time — so be careful with framing bits and units.

Quick check

1. A 2400-bit message is sent at 9600 bit/s. How long does it take?
2. What does an RTU do when the communication link fails?
3. Name two protocols that support time-stamped event reporting.
4. Why is report by exception preferred on slow radio links?

Answers: 1. 0.25 s. 2. It continues local logic and control, and buffers time-stamped data for later transmission. 3. DNP3 and IEC 60870-5-101/104 (also IEC 61850). 4. It sends only changed values, cutting traffic and delivering alarms faster than waiting for the next poll.