20.3 Debottlenecking, Optimization, and Troubleshooting

20.3 Debottlenecking, Optimization, and Troubleshooting

The NCEES specification lists Process improvement and troubleshooting (e.g., debottlenecking, optimization) as the last subtopic under Operation and Maintenance. These are the questions an operating-plant engineer actually gets asked, and they reward systematic method over recall.

---

1. Debottlenecking

A plant runs at the rate of its most constraining unit. Capacity analysis proceeds by expressing every unit's maximum sustainable rate as a percentage of current rate, then finding the minimum.

Worked example. A plant seeks a (20%) throughput increase. Unit capacities, evaluated as a percentage of current rate:

The binding constraint is the relief system at 115%, not the column at 118%. Every relief scenario — blocked outlet, fire case, loss of cooling, control valve failure — scales with the process rate, so raising throughput raises required relieving capacity and can push existing relief devices and the flare header past their rated capacity. This is the constraint most often missed, because it is not visible in the daily operating data; the flare system reveals its inadequacy only during the emergency it was installed to handle.

Debottlenecking therefore follows a disciplined loop, which is the Theory of Constraints applied to a process plant:

  1. IDENTIFY the constraint            (capacity analysis, not intuition)
  2. EXPLOIT it                          (eliminate its downtime, tighten its
                                          control, cut its recycle load - free capacity first)
  3. SUBORDINATE everything else         (do not run upstream units faster than
                                          the constraint can absorb)
  4. ELEVATE it                          (capital: add area, add trays, replace internals)
  5. REPEAT - the bottleneck has moved

Step 2 is where the return is. Before buying a bigger column, check whether reflux ratio is above its economic optimum (Section 19.2), whether the feed is entering at the wrong tray, whether an exchanger is fouled (Section 9.3), or whether a recycle stream is consuming capacity that a small purge change would free.

Typical constraints and their usual relief:

---

2. Heat Integration and Pinch Analysis

Energy is usually the largest controllable operating cost. Pinch analysis sets the thermodynamic minimum utility requirement before any exchanger network is designed.

Method (the problem table algorithm):

1. List every hot and cold stream with its supply and target temperatures and its heat capacity flow rate (CP).
2. Choose (\Delta T_{min}) (typically (10)-(20^\circ\text{C}); smaller means less utility but more area).
3. Shift temperatures: hot streams down by (\Delta T_{min}/2), cold streams up by (\Delta T_{min}/2).
4. Form temperature intervals at every shifted boundary and compute each interval's surplus or deficit as ((\sum CP_{hot} - \sum CP_{cold}) \Delta T).
5. Cascade the surpluses downward from the top. The most negative cumulative value is the minimum hot utility (Q_{H,min}); the temperature at which it occurs is the pinch.
6. (Q_{C,min} = Q_{H,min} + \sum Q_{hot} - \sum Q_{cold}).

Worked example. (\Delta T_{min} = 20^\circ\text{C}).

Shifted: H1 (240 \to 110); H2 (190 \to 90); C1 (100 \to 240); C2 (140 \to 220). Interval boundaries: (240, 220, 190, 140, 110, 100, 90).

Cascading from zero at the top: (-10,\ -115,\ -140,\ -65,\ -60,\ -30). The most negative value is (-140), so:

$Q_{H,min} = 140\text{ kW}, \qquad \text{pinch at shifted } 140^\circ\text{C} \Rightarrow \text{hot side } 150^\circ\text{C},\ \text{cold side } 130^\circ\text{C}$

$Q_{C,min} = 140 + 560 - 590 = 110\text{ kW}$

The three golden rules follow directly, because the pinch is the point of zero net heat flow:

1. Do not transfer heat across the pinch. Every kilowatt sent across it adds one kilowatt to both the hot and the cold utility.
2. No external cooling above the pinch.
3. No external heating below the pinch.

Violating any one of the three is the most common defect in an existing exchanger network, and finding a cross-pinch exchanger is often the single largest energy saving available in an operating plant.

---

3. Troubleshooting Method

Undisciplined troubleshooting changes several variables at once and learns nothing. The method:

1. Define the problem precisely. What changed, when, and how quickly? A step change points to an equipment or instrument event; a gradual drift points to fouling, catalyst decay, or corrosion.
2. Verify the measurement before believing it. A large share of apparent upsets are instrument faults. Check the DP-level transmitter against a sight glass, the orifice against a second flow indication, the thermocouple against a neighbor. Section 18.4 gives the two classic instrument errors: the square-root relation on an orifice DP, and the density dependence of DP level.
3. Close the balances. A material balance that no longer closes localizes the problem faster than any other single test: it shows whether material is leaving where it should not. An energy balance that no longer closes points to fouling, bypassing, or a leaking exchanger.
4. Hypothesize and rank by likelihood (\times) cheapness to test. Test the cheap likely one first.
5. Change one variable at a time, and allow the process to reach steady state before judging the result — many loops have time constants measured in hours.

Diagnostic signatures worth memorizing:

The single most useful troubleshooting habit is to ask what changed immediately before the symptom appeared — feedstock, ambient temperature, a maintenance activity, a setpoint, a catalyst load, a control loop returned to automatic. Processes rarely degrade spontaneously, and the change log is usually a faster diagnostic than the process data.