Process control and instrumentation

In short

Process control and instrumentation is the practice of measuring process variables (flow, temperature, pressure, level, composition), comparing them with desired set points, and manipulating final control elements - usually control valves - so a plant runs safely, at specification and at minimum cost. Instrumentation supplies the measurement and signalling hardware; control supplies the decision logic (on-off, P, PI, PID, cascade, feedforward, ratio, multivariable). It is a high-weight GATE-CH section (Instrumentation and Process Control is a named part of the GATE-CH syllabus) and the single most asked practical area in interviews for Reliance, IOCL, BPCL, HPCL, GAIL, ONGC, Tata Chemicals, UPL, Deepak Nitrite, Pidilite, Honeywell, Yokogawa, Emerson, ABB, L&T and EIL. Working engineers meet it as DCS operation, loop checking, instrument datasheets, alarm management and SIL-rated safety loops under the Petroleum & Natural Gas Rules and PESO-certified equipment requirements.

Explained

The feedback loop and its elements

• Four elements of every loop: sensor/transmitter (measures, converts to a standard signal) → controller (compares with set point, computes output) → final control element (valve, VFD, damper) → process (responds, changing the measured variable).
• Standard signals: 4-20 mA DC for analogue transmission (4 mA = 0% so a broken wire reads 0 mA and is detectable - "live zero"); 3-15 psig (0.2-1.0 kg/cm²g) pneumatic for valve actuators; 1-5 V DC inside panels; digital via HART (digital superimposed on 4-20 mA), Foundation Fieldbus, PROFIBUS-PA, Modbus.
• Error e(t) = set point − measured value. Direct-acting controller: output rises when measurement rises. Reverse-acting: output falls when measurement rises.
• Loop action rule: the product of the signs of process gain, valve action and controller action must be negative for negative feedback. Example: steam-heated reboiler with air-to-open steam valve → process gain positive → controller must be reverse-acting.
• Servo problem: set point changes, disturbance constant. Regulator problem: set point fixed, disturbance changes - most industrial loops are regulator problems.
• Measurement terms: accuracy (closeness to true value), precision/repeatability, span, range, zero, turndown ratio, resolution, hysteresis, dead band, drift, rangeability, sensitivity, response time (T63), damping.

Controller modes

• On-off (two-position): output is fully on or off with a differential gap; cycles continuously. Used for domestic thermostats, level switches, jacket heaters.
• Proportional (P): p = p̄ + Kc·e. Proportional band PB (%) = 100/Kc. Simple and fast, but leaves offset (steady-state error) for a sustained load change; offset shrinks as Kc rises but stability falls.
• Integral / reset (I): output changes at a rate proportional to error, so it integrates until e = 0 - eliminates offset. Penalty: adds phase lag, slows response, can reset-windup when the valve saturates (cured by anti-reset-windup / external reset feedback). Reset rate = 1/τI repeats per minute.
• Derivative / rate (D): acts on the rate of change of error, giving anticipatory action; adds stability and speeds recovery in sluggish loops with large thermal capacity. Never used alone (no action at steady error), amplifies noise, so it is avoided on noisy flow and pressure loops and usually filtered.
• Practical selection: flow and liquid pressure → PI with low gain (fast, noisy); level → P or PI with high gain (averaging level control deliberately uses loose tuning to absorb surges); temperature → PID; composition/analyser loops → PID with dead-time compensation or sampled control.
• Reverse/direct and bumpless transfer, auto-manual station, output tracking are standard DCS features examined in vivas.

Tuning methods

1. Ziegler-Nichols continuous cycling (closed loop): set I and D off, raise Kc until the loop oscillates with constant amplitude → ultimate gain Kcu and ultimate period Pu. Settings: P → Kc = 0.5Kcu; PI → Kc = 0.45Kcu, τI = Pu/1.2; PID → Kc = 0.6Kcu, τI = Pu/2, τD = Pu/8. Gives quarter-decay ratio, fairly aggressive.
2. Cohen-Coon (open loop / process reaction curve): apply a step to the valve in manual, fit the response to a first-order-plus-dead-time model Kp·e^−θs/(τs+1), then use the Cohen-Coon formulas; good for loops with θ/τ up to about 1.
3. IMC / lambda tuning: choose a desired closed-loop time constant λ; for FOPDT with PI, Kc = τ/[Kp(λ+θ)], τI = τ. Gives smooth, non-oscillatory control - the method most used in modern DCS practice.
4. Integral-criteria tuning: minimise IAE, ISE or ITAE (Lopez correlations).
5. Field "trial and error": halve gain, double reset until acceptable - acceptable only with a documented before/after trend.
6. Performance measures: rise time, settling time, overshoot, decay ratio (0.25 = quarter decay), offset, period of oscillation, IAE.
7. Stability: a loop is stable if all roots of 1 + GcGvGpGm = 0 lie in the left half-plane (Routh test); gain margin ≥ 1.7-2 and phase margin ≥ 30-45° are usual design targets. Dead time is the main destabiliser - every loop should be designed to minimise it (place sensors close, avoid long sample lines).

Advanced and multi-loop structures

• Cascade control: a fast secondary (slave) loop inside a slow primary (master) loop; the master's output is the slave's set point. Requirement: the secondary must be at least 3-5 times faster than the primary and must see the disturbance first. Classic examples - reactor temperature (master) cascaded to jacket-coolant flow (slave); furnace outlet temperature cascaded to fuel-gas pressure/flow.
• Feedforward control: measures the disturbance and acts before the controlled variable moves; needs a process model, cannot correct modelling error, so it is always combined with feedback trim. Used on feed-flow changes to distillation columns and heat-exchanger duty control.
• Ratio control: holds the ratio of two flows constant (air-to-fuel in furnaces, reactant ratio in reactors); implemented either by multiplying the wild stream by the ratio or dividing two measurements.
• Override / selective control: high or low signal selectors protect equipment (compressor suction pressure override, pump minimum-flow override).
• Split-range control: one controller output drives two valves over different parts of the range (0-50% heating, 50-100% cooling in a reactor jacket).
• Inferential control: a hard-to-measure variable (composition) is inferred from easy ones (tray temperature plus pressure-compensated temperature) - standard in distillation.
• Dead-time compensation: Smith predictor; internal model control.
• Multivariable / MPC: model predictive control (DMC, RMPCT) handles interaction and constraints in distillation, FCC and crude units; relative gain array (RGA) guides which manipulated-controlled pairing minimises interaction; decouplers cancel cross-effects.
• Batch and sequence control: PLC/sequential function charts, recipe management per ISA-88 (batch) and ISA-95 (plant-to-ERP integration).
• Control of unit operations: distillation (reflux ratio, boil-up, pressure, levels - degrees of freedom usually 5), CSTR (temperature by jacket, level by outflow), heat exchanger (bypass control gives faster response than throttling the utility), compressor (anti-surge control), pH (highly nonlinear titration curve, needs equal-percentage valve or nonlinear gain).

Measuring instruments

• Flow: differential-pressure types - orifice plate (cheap, rangeability only about 3:1-4:1, permanent loss high), venturi (low loss, costly), flow nozzle, pitot/averaging pitot; variable-area rotameter; positive displacement; turbine; electromagnetic (conductive liquids only, no obstruction, slurries OK); ultrasonic (clamp-on, transit-time or Doppler); vortex shedding; Coriolis (direct mass flow plus density, highest accuracy ±0.1%, costly); thermal mass for gases. Orifice: Q ∝ √ΔP, so square-root extraction is needed and accuracy collapses below ~25% of span.
• Temperature: thermocouples (Seebeck effect; Type K chromel-alumel up to ~1,260 °C, J, T, E, R/S/B for high temperature; cheap, wide range, needs cold-junction compensation, lower accuracy); RTD Pt100 (α = 0.00385 Ω/Ω/°C, −200 to 650 °C, best accuracy and stability; use 3- or 4-wire to cancel lead resistance); thermistor (very sensitive, nonlinear, narrow range); bimetallic and filled-system dial gauges; thermowell needed for any insertion into pressure-containing pipe (adds lag of 10-60 s); radiation/IR pyrometers and optical pyrometers for furnaces and kilns.
• Pressure: manometers, Bourdon tube, bellows, diaphragm; electronic transmitters using capacitance, piezoresistive strain-gauge or resonant-silicon sensing; absolute, gauge and differential types; chemical seals/diaphragm seals and capillary fill for corrosive or polymerising fluids; pressure gauges on vessels must have a gauge isolation valve.
• Level: sight glass; float and displacer; differential pressure with wet/dry leg (zero elevation and suppression calculations are examined); capacitance; ultrasonic and radar (non-contact) and guided-wave radar (a probe in contact with the product), now the default for most tanks; nucleonic gamma level (for very hostile service, needs AERB licence in India); load cells for weigh hoppers; interface level by displacer or GWR.
• Composition/analysers: gas chromatograph (large dead time due to cycle), NDIR CO/CO₂, zirconia oxygen analyser (combustion control), pH/conductivity, moisture, density, viscosity, opacity/CEMS stack analysers for CPCB continuous emission monitoring.
• Transmitter features: smart/digital transmitters with HART give remote range change, diagnostics and digital accuracy; two-wire loop power; damping and square-root extraction configurable; calibration = 5-point up/down check for zero, span, linearity and hysteresis against a reference standard traceable to NPL (National Physical Laboratory, India) through an NABL-accredited lab.

Final control elements

• Control valve parts: body (globe, butterfly, ball, plug, diaphragm), trim (plug, seat, cage), actuator (spring-diaphragm pneumatic, piston, electric, electro-hydraulic), positioner (improves accuracy, needed for large valves, high friction, split range or cascade), I/P converter, handwheel, limit switches, solenoid valve for trip service.
• Fail-safe action: air-to-open / fail-close (FC) and air-to-close / fail-open (FO) - chosen from process safety: fuel-gas valve fails closed, cooling-water valve fails open, reactor quench valve fails open.
• Inherent characteristics: linear (flow ∝ travel), equal-percentage (equal travel gives equal % change in flow - the usual choice when valve pressure drop varies, e.g. series with large pipe resistance), quick-opening (on-off service). Installed characteristic differs from inherent because pressure drop changes with flow; design for valve ΔP of 25-33% of total system drop at design flow to retain controllability.
• Valve sizing: Cv (gpm of water at 1 psi drop) or Kv (m³/h at 1 bar); cavitation when local pressure falls below vapour pressure (use anti-cavitation trim, multistage), flashing, choked flow; noise limits typically 85 dBA at 1 m.
• Other final elements: variable frequency drives on pumps and fans (energy saving over throttling), dampers, metering/dosing pumps, on-off/shut-off valves.

Plant control systems, safety and Indian practice

• DCS (Distributed Control System - Honeywell Experion, Yokogawa CENTUM, Emerson DeltaV, ABB 800xA, Siemens PCS7) for continuous analogue-heavy plants: controllers in field cabinets, operator HMI, historian, redundant networks. PLC for fast discrete logic and packages; SCADA for wide-area supervision (pipelines, water utilities); ESD/SIS (Emergency Shutdown / Safety Instrumented System) is physically and logically separate from the DCS.
• Functional safety: IEC 61508 / IEC 61511 (adopted in India as IS/IEC 61511) define the safety lifecycle; SIL 1-4 from required risk reduction; SIF components rated by PFDavg; proof-test intervals; 1oo2/2oo3 voting; LOPA sets the SIL target. Typical process plant SIFs are SIL 1 or SIL 2.
• Hazardous-area instrumentation: area classification into Zone 0/1/2 (gases) per IS/IEC 60079-10; protection concepts - Ex ia/ib intrinsic safety (energy limited, barriers/isolators), Ex d flameproof (IS/IEC 60079-1; IS 2148 covers flameproof enclosures), Ex e increased safety, Ex p purged. Gas groups IIA/IIB/IIC, temperature classes T1-T6. Equipment for use in hazardous areas in India requires PESO/CCOE approval and a BIS/Indian certifying-body certificate; electrical installation follows the Central Electricity Authority (Measures relating to Safety and Electric Supply) Regulations and IS 732 wiring practice.
• Other codes and practice: instrument symbols and tag numbers per ISA-5.1 (adopted in Indian P&ID practice); alarm management per ISA-18.2 (and the EEMUA guidance); environmental monitoring - CPCB-mandated online continuous emission and effluent monitoring with data telemetry; occupational safety obligations now sit under the Occupational Safety, Health and Working Conditions Code, 2020, in force from 21 November 2025, which replaced the Factories Act, 1948; MSIHC Rules obligations for major accident hazard installations; relief devices under IS 4231 / API 520-521 sizing practice (a relief valve is protection, not control). Weights and measures instruments used in custody transfer need Legal Metrology approval.
• Engineering deliverables a fresher handles: instrument index, specification sheets/datasheets, I/O list, loop diagrams, hook-up drawings, cause-and-effect matrix, cable schedule, junction-box layout, loop folder, calibration certificates, FAT/SAT, pre-commissioning loop checking and interlock testing.

Key points

• Feedback loop - sensor/transmitter, controller, final control element and process in a closed path; corrective action begins only after the error appears.
• 4-20 mA live zero - a 0 mA reading means a broken loop, not 0% process; 3-15 psig is the pneumatic equivalent.
• Proportional offset - P-only control always leaves a steady-state error for a sustained load change; integral action removes it.
• Reset windup - integral keeps integrating while the valve is saturated; anti-windup or external reset feedback is needed.
• Derivative action - anticipatory, stabilising, never used alone, unsuitable for noisy flow and pressure loops.
• Proportional band PB% = 100/Kc; high gain = narrow band = tighter but less stable control.
• Ziegler-Nichols - Kcu and Pu from sustained oscillation; PID → Kc = 0.6Kcu, τI = Pu/2, τD = Pu/8.
• Dead time is the most destabilising element in a loop; minimise it by sensor placement, not by tuning.
• Cascade - fast inner loop takes the outer loop's output as set point; needs clear time-scale separation.
• Feedforward - acts on the measured disturbance before the controlled variable moves; always trimmed by feedback.
• Ratio control - keeps two flows in fixed proportion (air-to-fuel, reactant feed).
• Orifice plate - Q ∝ √ΔP, rangeability ~3:1, high permanent loss; Coriolis gives direct mass flow at ±0.1%.
• RTD vs thermocouple - Pt100 is accurate and stable but limited to about 650 °C; thermocouples are cheap and go far hotter but need cold-junction compensation.
• Valve characteristic - equal-percentage for varying ΔP (most process service), linear when valve drop is nearly constant, quick-opening for on-off.
• Valve ΔP allocation - give the valve 25-33% of the total dynamic pressure drop at design flow for controllability.
• Fail-safe choice - fail-close for fuel and feed, fail-open for cooling and quench; decided by process hazard, not convenience.
• DCS vs PLC vs SIS - DCS for continuous regulatory control, PLC for fast discrete logic, SIS independent and SIL-rated to IEC 61511.
• Hazardous area - Zone 0/1/2, gas groups IIA/IIB/IIC, intrinsic safety (Ex ia) or flameproof (Ex d, IS 2148); PESO approval required in India.

Formulas and examples

Example 1 - 4-20 mA transmitter scaling. A pressure transmitter is ranged 0-25 bar g and reads 13.6 mA. Also find the current at 18 bar.

• Percent = (13.6 − 4)/16 = 9.6/16 = 0.60 = 60% → PV = 0.60 × 25 = 15.0 bar g.
• At 18 bar: fraction = 18/25 = 0.72 → I = 4 + 16 × 0.72 = 15.52 mA.

Example 2 - proportional offset and proportional band. A level process has Kp = 2 m per unit valve signal, Kv = 1, Km = 1 (measurement in m). A P-only controller has Kc = 4. A load disturbance that would raise level by 0.5 m at steady state (Kd·d = 0.5 m) occurs.

• Loop gain K = KcKvKpKm = 4 × 1 × 2 × 1 = 8.
• Offset = 0.5/(1 + 8) = 0.5/9 = 0.0556 m ≈ 5.6 cm of permanent error.
• With Kc = 20: offset = 0.5/(1 + 40) = 0.0122 m ≈ 1.2 cm - smaller, but PB narrows from 100/4 = 25% to 100/20 = 5%, so stability margin falls.
• Adding integral action drives the offset to zero at steady state regardless of Kc.

Example 3 - Ziegler-Nichols and IMC tuning compared. A temperature loop oscillates with constant amplitude at Kc = 5.0, period 12 min. An open-loop step test fits Kp = 0.8 °C/%, τ = 10 min, θ = 2 min.

• Z-N PID: Kc = 0.6 × 5.0 = 3.0, τI = 12/2 = 6 min, τD = 12/8 = 1.5 min.
• Z-N PI: Kc = 0.45 × 5.0 = 2.25, τI = 12/1.2 = 10 min.
• IMC-PI with λ = θ = 2 min: Kc = 10/[0.8 × (2 + 2)] = 10/3.2 = 3.125 %/°C, τI = 10 min.
• IMC-PI with a conservative λ = 3τ... instead take λ = 10 min: Kc = 10/[0.8 × 12] = 1.04, τI = 10 min - much gentler, little overshoot. Larger λ = smoother and slower; smaller λ = tighter and more oscillatory.

Example 4 - control valve sizing and ΔP check. Water (SG = 1.0) at 40 m³/h must pass a control valve with 1.4 bar drop across it; total pump-to-vessel dynamic drop is 4.5 bar.

• 40 m³/h = 40 × 4.403 = 176.1 gpm; ΔPv = 1.4 bar = 1.4 × 14.5 = 20.3 psi.
• Cv = Q/√(ΔPv/SG) = 176.1/√20.3 = 176.1/4.506 = 39.1 → select a valve with rated Cv ≈ 55-60 so design flow sits near 65-70% travel.
• Valve share of drop = 1.4/4.5 = 31% - inside the recommended 25-33%, so the installed characteristic will stay reasonably linear with an equal-percentage trim.
• Kv = 0.865 × 39.1 ≈ 33.8.

Example 5 - Pt100 and orifice square-root. (a) A Pt100 reads 157.3 Ω. (b) An orifice-DP transmitter is calibrated 0-250 mbar for 0-100 m³/h; it reads 90 mbar.

• (a) t = (157.3 − 100)/(100 × 0.00385) = 57.3/0.385 = 148.8 °C.
• (b) Flow = 100 × √(90/250) = 100 × √0.36 = 100 × 0.60 = 60 m³/h. Note that 36% of DP span gives 60% of flow - which is why DP flow loops need square-root extraction and lose accuracy at low flow.

Where it is used

• Refineries and petrochemicals: Reliance Jamnagar, IOCL Panipat/Paradip, BPCL Kochi, HPCL Visakh - DCS-based crude, FCC and reformer control, APC/MPC layers, SIS for furnaces and compressors; roles: panel/DCS engineer, APC engineer, instrumentation maintenance.
• Fertiliser, chlor-alkali and bulk chemicals: IFFCO, Coromandel, GSFC, Tata Chemicals, Grasim, Aditya Birla Chemicals - ammonia-syngas ratio control, reactor temperature cascades, pH and chlorine analyser loops.
• Pharma and specialty chemicals: Sun Pharma, Cipla, Divi's, Aarti, Deepak Nitrite - batch PLC/recipe control, 21 CFR Part 11-compliant data logging, CIP/SIP sequences, calibration and qualification (IQ/OQ/PQ) documentation.
• Power, cement and steel: NTPC, Tata Steel, UltraTech - boiler drum level three-element control, combustion air-fuel ratio, kiln and mill control, CEMS reporting to CPCB/SPCB.
• EPC and consultancy design roles: EIL, L&T Hydrocarbon, Toyo, Tecnimont, Thyssen - instrument datasheets, I/O lists, loop and hook-up drawings, cause-and-effect matrices, SIL verification.
• Instrument vendors and systems integrators: Honeywell, Yokogawa, Emerson, ABB, Siemens, Endress+Hauser, Forbes Marshall - sales/application engineering, commissioning, FAT/SAT, service.

Interview questions

• Q: Why is the industrial analogue standard 4-20 mA rather than 0-20 mA? A: The 4 mA live zero lets the system distinguish a genuine 0% reading from a broken wire or dead transmitter, and the residual current can power a two-wire transmitter.
• Q: What does integral action do that proportional action cannot? A: It keeps changing the output as long as any error remains, so it removes the steady-state offset that P-only control leaves after a sustained load change.
• Q: Why is derivative action avoided on flow loops? A: Flow measurements are noisy and flow processes are fast; derivative amplifies the noise into violent valve movement without any stability benefit.
• Q: How do you decide whether a controller should be direct or reverse acting? A: Make the product of the process, valve and controller gain signs negative - e.g. with an air-to-open steam valve on a heater (positive process gain) the controller must be reverse acting.
• Q: When is cascade control justified? A: When a measurable intermediate variable responds much faster (3-5 times) than the primary variable and sees the disturbance first - such as jacket coolant flow inside a reactor temperature loop.
• Q: Why must feedforward always be combined with feedback? A: Feedforward relies on a process model and on measuring only the known disturbance; feedback trims the residual error from model inaccuracy and unmeasured upsets.
• Q: Why choose an equal-percentage valve? A: When the valve's share of pressure drop falls as flow rises (series with high line resistance), the equal-percentage inherent characteristic compensates and gives a near-linear installed characteristic and constant loop gain.
• Q: What fraction of the system pressure drop should the control valve take, and why? A: Roughly 25-33% at design flow - too little and the valve loses authority and rangeability, too much wastes pumping power.
• Q: What is reset windup and how do you prevent it? A: When the valve saturates, integral keeps accumulating error so the output overshoots badly on recovery; it is prevented by anti-reset-windup limits, external reset feedback or clamping the integral.
• Q: An orifice flow transmitter reads 50% of DP span. What is the flow? A: √0.5 = 70.7% of maximum flow - DP is proportional to the square of flow, so square-root extraction is essential.
• Q: How is a safety instrumented system different from the DCS loop? A: The SIS is an independent, SIL-rated protection layer per IEC 61511 with its own sensors, logic solver and final elements, proof-tested at fixed intervals; the DCS is the normal control layer and is not credited as protection.
• Q: What is the difference between intrinsic safety and flameproof protection? A: Intrinsic safety (Ex ia/ib) limits energy in the circuit so it can never ignite the atmosphere; flameproof (Ex d, IS 2148) allows ignition inside a robust enclosure but contains and cools the flame so it cannot propagate outside.
• Q: Why does a thermowell slow down a temperature loop? A: It adds thermal mass and an air/oil gap, introducing a lag of tens of seconds, which effectively acts as extra dead time and reduces achievable loop performance.

Common mistakes

• Confusing proportional band with gain, or quoting Kc without stating whether signals are in % or engineering units.
• Writing that integral action removes offset "instantly" - it removes steady-state offset but slows the response and reduces stability margin.
• Forgetting square-root extraction on orifice/DP flow, or using a DP flowmeter below about 25% of span where its accuracy collapses.
• Choosing fail-safe valve action by habit instead of by hazard analysis - a fuel valve left fail-open is a serious design error.
• Sizing a control valve on the full pump head instead of the pressure drop actually available across the valve, and ignoring cavitation and choked flow.
• Tuning a loop to remove oscillation when the real cause is excessive dead time, a sticking valve without a positioner, or a wrongly placed sensor.
• Treating the DCS alarm or control loop as a safety layer, and omitting the independent SIS, voting and proof-test interval required by IEC 61511.
• Using a 2-wire Pt100 over a long cable run and blaming the resulting positive error on the sensor instead of lead resistance.

Related topics

Process dynamics and transfer functions · Process flow diagrams and P&IDs · Process plant safety: HAZOP and PSM · Process design and optimisation · Fire, explosion and toxic release hazards · Plant layout and piping