Case study: recommended changes and convincing

Assumed case study: a milk pasteurisation and packaging section of a dairy plant (HTST process: raw milk tank, heat exchanger, steam-heated holding tube, cooling, filling). The existing system uses analog panel indicators, bimetal thermostats, a manual steam valve and operator logging. (If your own visit was to another plant, replace the process names; the structure of the answer stays the same.)

Recommended changes

1. Replace manual steam-valve control by a closed-loop microcontroller/PID temperature controller with a PT100 sensor.
2. Add level sensors with automatic pump control, a flow-diversion valve, and alarms (over-temperature, low level, pump fault).
3. Provide data logging and an RS-485/Modbus link to a PC for supervision and reports.
4. Add a UPS, surge protection, proper grounding and shielding, and a manual bypass.

Why management should implement it

• Quality: consistent pasteurisation, fewer rejected batches, and records for food-safety audits.
• Cost: less steam waste and labour; payback in about 10 months (see below).
• Safety and reliability: alarms and interlocks reduce accidents and breakdowns.

Cost-benefit (assumed figures): investment Rs 12,00,000 (controller panel, sensors, valves, SCADA PC, installation, training). Monthly savings: manpower Rs 60,000, steam/energy Rs 35,000, reduced rejection Rs 20,000 = Rs 1,15,000. Payback = 12,00,000 / 1,15,000 ≈ 10.4 months; yearly return = 13.8 lakh on 12 lakh, about 115 percent.

Convincing the senior engineer technically

Present the technical design and justify each choice:

 PT100 --> Amp/SigCond --+
 Level sensor ----------+--> MUX/ADC (12-bit)
 Flow sensor ---(pulse)-+         |
 Keypad ----------------+---> [Microcontroller] <--> LCD 16x2
                               |  ARM/AVR + flash   |
   Opto + SSR/relay <----------+  watchdog, timers  +--> RS-485
     |                         |                          (Modbus)
 Pump / steam valve / alarm    +--> EEPROM log         PC / SCADA
 Power supply: 24 V + 5 V + UPS

• Choice of controller: a microcontroller is enough because there are few I/O points, and it gives low cost, low power and deterministic timing; a microprocessor would add unnecessary complexity.
• Accuracy: a 12-bit ADC and an amplified PT100 give 0.037 C per count (calculated), well within the ±0.5 C requirement; use a 3-wire or 4-wire connection to cancel the lead resistance.
• Control algorithm: PID or hysteresis control with set-point limits; simulate and test on a bench model before installation.
• Reliability: watchdog timer, brown-out reset, isolation (opto/relay), TVS and filters, a redundant temperature sensor, a fail-safe valve (closes on power loss), and a manual mode.
• Interfacing: signal conditioning, ADC data format (12-bit, right justified), RS-485 at 9600 baud, 8N1, polling by the master with CRC.
• Debate points: a microcontroller is cheaper than a PLC but less standard (answer: documented design, spares and calibration procedure); noise immunity (shielded twisted pair, grounding); and maintainability (modular software, diagnostics).

Probable problems after implementation

Noise and ground loops, sensor calibration drift, operator training, firmware bugs, power quality, spare parts, and communication faults. Mitigation: a pilot run in parallel with the old system, a test plan, a training schedule, and a maintenance contract.