# Reactor Thermal Control (TCU) Troubleshooting: Eliminating Sluggish Cycles, Temperature Overshoots, and Valve Hunting

# A Quantitative Engineering Guide to Master-Slave Cascade PID Dynamics, 3-Way Valve Seat Leakage, Thermal Fluid Vapor Locking, and Heat Transfer Lag in Glass-Lined Reacting Systems

# Executive Summary

In batch pharmaceutical synthesis and fine chemical manufacturing, reactor temperature control is the master variable that governs reaction selectivity, byproduct impurity generation, crystallization polymorph stability, and process safety. Modern API facilities rely heavily on automated Temperature Control Units (TCUs / Mono-Fluid Systems) capable of ramping a single circulating thermal fluid (e.g., Syltherm, Marlotherm, Therminol) across temperatures from $-30^\circ\text{C}$ to $+200^\circ\text{C}$.

However, when a reactor temperature loop behaves erratically, the symptoms are unmistakable and costly:

• Severe Temperature Overshoots ($>5\text{–}15^\circ\text{C}$): Exothermic reactions overshoot their setpoint during reagent addition, triggering Stoessel criticality alarms, solvent boil-over, or product decomposition.
• Continuous Valve Hunting & Oscillation: Modulating 3-way valves cycle continuously between $10\%$ and $90\%$ stroke with a 15- to 45-minute period, causing mechanical actuator wear and immense utility steam/chilled brine waste.
• Sluggish Thermal Response: Heating or cooling ramps take 3 to 4 times longer than design calculations, stretching batch cycle times and tying up multi-million-dollar reactor assets.

Plant operations teams frequently blame the Distributed Control System (DCS) software or demand that automation engineers "retune the PID loop." In reality, over $80\%$ of reactor thermal control failures stem from underlying thermo-hydraulic anomalies: internal 3-way valve seat bypass leakage, air/vapor pockets trapped in jacket coils, thermal fluid degradation, or failure to account for the heavy thermal lag of glass-lined steel.

This technical engineering guide provides the mathematical cascade heat-transfer models, mechanical diagnostic procedures, cascade PID tuning algorithms, and a fully worked industrial troubleshooting case study for a 6,000 L commercial reacting system.

# 1. Mono-Fluid TCU Architecture & Cascade Control Dynamics

Unlike legacy batch plants that alternated between charging raw steam, cooling tower water, and chilled glycol directly into the reactor jacket—causing catastrophic thermal shock and rapid pipe scaling—modern facilities use a closed-loop mono-fluid system:

 ┌────────────────────────────────────────────────────────────────────────┐
 │                      TEMPERATURE CONTROL UNIT (TCU)                    │
 │                                                                        │
 │  Steam Heat Exchanger       Primary Circulating Loop      Chilled Brine│
 │  ┌─────────────────┐       ┌──────────────────────┐       ┌───────────┐│
 │  │ Steam Plate HEX │◄─────►│ High-Velocity Pump   │◄─────►│ Brine HEX ││
 │  └─────────────────┘       │ (Re > 50,000 in Coil)│       └───────────┘│
 │           ▲                └──────────┬───────────┘             ▲      │
 │           │ 3-Way Hot Valve           │                         │      │
 └───────────┼───────────────────────────┼─────────────────────────┼──────┘
             │                           ▼ Jacket Supply (T_jacket)│ 3-Way Cold Valve
       ┌─────┴───────────────────────────┴─────────────────────────┴─────┐
       │             COMMERCIAL GLASS-LINED REACTOR (GLR)                │
       │                                                                 │
       │   Process Fluid: T_process (Master Controlled Variable)         │
       │   Jacket Coil:   T_jacket  (Slave Controlled Variable)          │
       └─────────────────────────────────────────────────────────────────┘

# 1.1 The Master-Slave Cascade PID Architecture

Controlling batch temperature via a single PID loop that modulates utility valves directly from process fluid temperature ($T_{process}$) is mathematically unstable due to the massive thermal capacitance of the vessel:

1. Master Controller (Process Fluid Loop):
   • Compares measured reactor temperature ($T_{process}$) against the operator setpoint ($T_{set}$).
   • Outputs a calculated target jacket temperature ($T_{jacket,set}$).
   • Runs with a slow integral time ($T_i \approx 15\text{–}30\text{ min}$) to prevent reset windup.
2. Slave Controller (Jacket Loop):
   • Compares circulating jacket inlet temperature ($T_{jacket,in}$) against the target set by the Master ($T_{jacket,set}$).
   • Modulates the fast-acting 3-way heating and cooling valves.
   • Runs with high proportional gain and fast response ($T_i \approx 30\text{–}60\text{ seconds}$), rejecting utility pressure/temperature disturbances before they penetrate the reactor wall.

T_set ──► [ MASTER PID ] ──► T_jacket,set ──► [ SLAVE PID ] ──► Valve Position (%) ──► Reactor Jacket
              ▲                                   ▲
              │                                   │
       T_process (Sensor)                  T_jacket,in (Sensor)

# 2. Thermal Resistances in Series & The Glass-Lined Thermal Lag

In glass-lined steel reactors (GLR), heat transfer between the circulating jacket fluid and the agitated batch must traverse five thermal resistances in series:

Where:

• $U$: Overall heat transfer coefficient ($\text{W/m}^2\cdot\text{K}$)
• $h_i$: Process-side agitated liquid film coefficient ($\approx 400\text{–}800\text{ W/m}^2\cdot\text{K}$)
• $x_{glass} / k_{glass}$: Thermal resistance of the borosilicate glass lining ($x_{glass} \approx 1.5\text{–}2.0\text{ mm}$, $k_{glass} \approx 1.2\text{ W/m}\cdot\text{K}$)
• $x_{steel} / k_{steel}$: Thermal resistance of the carbon steel shell ($x_{steel} \approx 20\text{–}35\text{ mm}$, $k_{steel} \approx 45\text{ W/m}\cdot\text{K}$)
• $h_j$: Jacket-side forced-convection film coefficient ($\approx 1,200\text{–}2,500\text{ W/m}^2\cdot\text{K}$)
• $R_{fouling}$: Internal jacket scale and process fouling factor ($\text{m}^2\cdot\text{K/W}$)

Process Fluid (hi) ──► Glass Lining (1.2 W/m·K) ──► Steel Shell (45 W/m·K) ──► Fouling ──► Jacket Fluid (hj)
[High Resistance]        [DOMINANT RESISTANCE!]      [Low Resistance]                       [Low Resistance]

# 2.1 The Deadly Thermal Lag Time Constant ($\tau_d$)

Because the glass lining has a thermal conductivity ($1.2\text{ W/m}\cdot\text{K}$) that is $37\times$ lower than carbon steel, a step-change in jacket fluid temperature requires several minutes to begin influencing the process fluid:

# 3. The 4 Root Causes of Severe TCU Dysfunction

# 3.1 Fault 1: 3-Way Modulating Valve Seat Bypass Leakage

The most pervasive hardware failure in mono-fluid skids is internal valve passing. Over time, thermal cycling, particulate grit, and actuator calibration drift prevent modulating 3-way control valves from seating completely ($>99.5\%$ shutoff).

                      [ Hot Oil Circuit: 180°C ]
                                  │
                                  ▼
      3-Way Heating Valve Passing Just 3% Across Closed Seat!
                                  │
                                  ▼
[ Circulating Loop: Target -10°C Cooling Ramp ] ◄── Continuous Heat Bleed!

• Physical Consequence: When the TCU is commanded to cool the batch, the cooling valve opens. However, if the heating valve is leaking just $3\text{ to }5\%$ of its hot flow across the closed port, the chilled brine must continuously fight against incoming steam heat.
• Diagnostic Test: Touch or infrared-scan the pipe downstream of the closed utility valve. If the pipe downstream of the closed steam heat exchanger is warm ($>40^\circ\text{C}$) during a deep chilling cycle, the valve seat is passing.

# 3.2 Fault 2: Thermal Fluid Vapor Locking & Air Trapping

When a reactor jacket or half-pipe coil is drained and refilled, air pockets naturally congregate at high-point bends:

 Reactor Half-Pipe Coil:
 ┌──────────┐      Air Pocket Trapped at Coil Crown!
 │          │ ◄─── (Zero Heat Transfer: k_air = 0.026 W/m·K)
 │  Fluid   │
 └──────────┘

• Physical Consequence: Trapped air or thermal fluid light ends (flashed volatiles) form non-conductive vapor cushions. Wetted heat transfer area ($A$) drops by $30\text{ to }50\%$. The circulating pump cavitates intermittently, flow rate fluctuates, and temperature response turns sluggish.
• Diagnostic Test: Check the differential pressure gauge ($\Delta P$) across the reactor jacket supply and return. If $\Delta P$ is lower than the pump design curve or vibrates erratically, vapor is trapped in the coils. Perform high-velocity fluid purging with the jacket vent valves open to the expansion tank.

# 3.3 Fault 3: Agitator Baffle Cavitation & Sensor Sheath Detachment

In batch reactors, process temperature is measured by an RTD (Pt-100) embedded inside a glass-coated beaver-tail or finger baffle tip.

• If the thermal response time of the RTD assembly exceeds $30\text{ seconds}$ (due to missing thermal conductive paste inside the thermo-well), the controller receives lagging temperature data.
• If the batch liquid level drops below the RTD immersion depth during solvent boil-down, the sensor measures headspace vapor rather than reacting slurry, causing the TCU to overheat the batch.

# 3.4 Fault 4: Controller Reset Windup During Reagent Addition

During semi-batch reagent dosing, the chemical exotherm ($\dot{Q}_{rxn} = \Delta H_{rxn} \cdot \dot{n}_{feed}$) rapidly injects heat into the system. If the master controller's integral action is allowed to accumulate error while the jacket is already at maximum cooling ($T_{jacket} = T_{min}$), integral windup occurs. When dosing stops, the jacket remains locked at extreme cooling, freezing the batch.

# 4. Cascade Tuning Rules for Glass-Lined Reacting Vessels

Tuning a jacketed reactor requires a structured, two-step hierarchical sequence: tune the slave loop first, then tune the master loop.

Step 1: Place Master in MANUAL ──► Tune SLAVE Loop (Jacket Fluid Response)
                                              │
                                              ▼ SLAVE Fast & Stable (Zero Overshoot)
Step 2: Place Master in AUTO   ──► Tune MASTER Loop (Reactor Process Fluid)

# 4.1 Implementing Dynamic Jacket Setpoint Clamping

To prevent thermal shock to glass lining ($|\Delta T_{shock}| \le 100^\circ\text{C}$) and eliminate extreme overshoots, configure dynamic output clamping on the Master controller:

For a standard 6,000 L GLR, set $\Delta T_{max} = 30^\circ\text{C}$ during normal operation. This restricts the jacket from ever pulling more than $30^\circ\text{C}$ below the actual batch temperature, completely eliminating the thermal inversion overshoot cycle.

# 5. Worked Industrial Case Study: Eliminating a 14°C Exothermic Overshoot in a 6,000 L GLR

# 5.1 Process Background & Failure Symptoms

• Equipment: 6,000 L Glass-Lined Reactor (GLR) with half-pipe jacket ($A = 18.5\text{ m}^2$), TCU mono-fluid system running Marlotherm SH thermal fluid.
• Process Operation: Fast, exothermic amine acylation reaction carried out at $+15^\circ\text{C}$.
• Reagent Dosing: Acid chloride dosed over 2.5 hours ($\Delta H_{rxn} = -145\text{ kJ/mol}$, peak heat generation $\dot{Q}_{rxn} = 85\text{ kW}$).
• The Failure:
  • As dosing commenced, the batch temperature jumped from $+15^\circ\text{C}$ to $+29.4^\circ\text{C}$ ($14.4^\circ\text{C}$ overshoot).
  • The high temperature triggered an emergency stop interlock, halting reagent feed.
  • The jacket plunged to $-20^\circ\text{C}$. The batch then plummeted to $+2.1^\circ\text{C}$ (undershoot).
  • The TCU entered a violent 45-minute limit-cycle oscillation, and dimer impurity spiked to $3.8\%$ (spec $\le 0.5\%$).

Setpoint: +15°C
Batch Temperature Trajectory (Failed):
+15°C ──(Dosing Starts)──► SPIKES TO +29.4°C! ──(Feed Halts)──► PLUNGES TO +2.1°C ──► Continuous 45-min Hunting!

# 5.2 Forensic Diagnostic Audit

# Audit Item 1: Infrared Thermography on 3-Way Valves

Thermal imaging revealed that during the cooling phase, the pipe downstream of the 3-way steam heat exchanger was running at $78^\circ\text{C}$, despite the valve commanding $0\%$ heating:

• Finding: The hot 3-way control valve had developed mechanical seat cavitation damage. $6.5\%$ of hot fluid ($165^\circ\text{C}$) was continuously leaking into the circulating loop, imposing a parasitic $28\text{ kW}$ heating load against the chilled brine chiller.

# Audit Item 2: Jacket Coil Flow & Pressure Drop

• Design jacket flow rate: $45\text{ m}^3/\text{h}$ ($\Delta P = 2.4\text{ bar}$).
• Measured jacket flow rate: $24.2\text{ m}^3/\text{h}$ ($\Delta P = 1.1\text{ bar}$).
• Finding: Opening the top coil vent released a continuous discharge of entrained air and degraded thermal fluid light ends. Two complete half-pipe circuits had been vapor-locked.

# Audit Item 3: Cascade PID Configuration Review

• The Master controller had no output limits: it commanded $T_{jacket} = -25^\circ\text{C}$ as soon as batch temp rose $+1^\circ\text{C}$.
• Derivative action ($T_d$) on the Master controller was set to $0\text{ s}$ (Disabled), leaving the loop blind to the rate-of-change of the chemical exotherm.

# 5.3 The 4-Step Engineering Remediation

1. Valve Overhaul & Tight Shutoff Seating: Replaced the damaged 3-way valve with a high-performance Class VI tight shutoff modulating globe valve with Kalrez seat inserts, reducing bypass leakage from $6.5\%$ to $<0.02\%$.
2. De-Aeration & Thermal Fluid Replacement: Performed full high-velocity solvent flushing and refilled with fresh, dry synthetic thermal fluid. Installed an automated continuous air/gas separator on the TCU expansion tank return line, restoring jacket flow from $24.2\text{ m}^3/\text{h}$ to $46.0\text{ m}^3/\text{h}$ ($h_j$ increased from $750$ to $1,850\text{ W/m}^2\cdot\text{K}$).
3. Master-Slave Cascade Re-Tuning:
   • Slave (Jacket Loop): $K_p = 3.2$, $T_i = 45\text{ s}$, $T_d = 0\text{ s}$.
   • Master (Reactor Loop): $K_p = 1.8$, $T_i = 18\text{ min}$, $T_d = 75\text{ s}$ (with $15\text{ s}$ derivative filter).
4. Dynamic Output Clamping & Feed-Forward Dosing Interlock:
   • Programmed dynamic jacket clamping: $T_{jacket}$ is restricted to $T_{process} \pm 20^\circ\text{C}$.
   • Added feed-forward anticipation: starting the dosing pump automatically pre-chills the jacket by $4^\circ\text{C}$ two minutes before acid chloride reaches the vessel.

# 5.4 Operational Results on Next Commercial Batch

# 6. Reactor TCU Plant Troubleshooting Checklist

Symptom: Temperature Overshoots / Erratic Hunting
│
├─► Check 1: 3-Way Valve Bypass Leakage Test
│     ├─► Run cooling ramp with heating commanded to 0%.
│     └─► Measure pipe temp downstream of hot heat exchanger:
│         └─► If pipe is warm (>40°C) → VALVE IS PASSING! Overhaul seat seals immediately.
│
├─► Check 2: Jacket Hydraulic Flow & Entrained Vapor
│     ├─► Check pump differential pressure (ΔP) across jacket:
│     │   └─► If ΔP < design or vibrating → Air / vapor pockets in half-pipe coils.
│     └─► ACTION: Open top coil manual bleed valves into expansion tank during high-flow circulation.
│
├─► Check 3: Cascade Controller Architecture
│     ├─► Is the loop running in Single-Loop mode?
│     │   └─► ACTION: Convert immediately to Master-Slave Cascade PID.
│     └─► Is the Master controller commanding extreme jacket swings?
│         └─► ACTION: Program Dynamic Output Clamping (T_jacket = T_process ± 20°C).
│
└─► Check 4: Sensor Integrity & Immersion Depth
      ├─► Has batch volume dropped below RTD baffle tip during distillation?
      └─► Verify thermal conductive paste is present inside RTD thermowell (Response time < 15 s).