CSTR vs PFR: Why Reactor Shape Changes the Volume You Need

Two Ideal Reactors, Two Opposite Mixing Assumptions

The continuous stirred-tank reactor and the plug flow reactor are the two idealised extremes of continuous reactor design. A CSTR is assumed perfectly mixed: the composition is the same everywhere inside it, and therefore equal to the composition of the outlet stream. A PFR is assumed to have no mixing at all in the flow direction: fluid moves through a tube as a series of thin slices, each reacting as it travels, like a batch reactor on a conveyor belt.

These assumptions lead to different design equations. For a CSTR, a mole balance on the whole vessel gives V = F_A0·X/(−r_A), with the rate evaluated at the exit conversion, because the whole tank sits at exit conditions. For a PFR, the balance is written on a thin slice and integrated along the length: V = F_A0 ∫ dX/(−r_A), from zero to the exit conversion.

The Levenspiel Plot: Rectangle vs Area

Octave Levenspiel's trick was to plot 1/(−r_A) against conversion. On that plot, the PFR's V/F_A0 is simply the area under the curve up to the exit conversion. The CSTR's V/F_A0 is a rectangle: the exit conversion times the curve's height at the exit. Both reactors can be read off the same chart.

For any ordinary positive-order reaction, the rate falls as reactant is consumed, so 1/(−r_A) rises with conversion and the curve slopes upward. The rectangle, whose height is the curve's highest point, must then enclose more area than the curve itself, and the CSTR needs more volume. The underlying reason is that the CSTR runs entirely at the low, exit concentration, while the PFR passes through every higher concentration on the way there.

How Big the Difference Gets

For a first-order reaction the difference can be worked out exactly. The CSTR needs kτ = X/(1 − X), while the PFR needs kτ = −ln(1 − X), where τ is the space time. At 50% conversion the CSTR needs 1.0/k against 0.69/k for the PFR, about 1.4 times the volume. At 90% conversion the gap widens to 9/k against 2.3/k, nearly four times the volume. The higher the required conversion, the worse the CSTR looks.

The comparison flips for reactions whose rate rises as conversion increases, such as autocatalytic reactions, where a product speeds up the reaction, and for some adiabatic exothermic reactions, where rising temperature outweighs falling concentration early on. Their Levenspiel curves dip before rising, and the smallest total volume comes from a CSTR operating at the curve's minimum followed by a PFR for the remainder.

Tanks in Series and Real Reactors

Splitting one big CSTR into several smaller ones in series closes most of the gap. Each tank operates at a progressively lower concentration, so on the Levenspiel plot the single large rectangle becomes a staircase of smaller ones that hugs the curve more closely. As the number of tanks grows, the total volume approaches that of a PFR, which is why a series of stirred tanks is a common practical substitute for a long tubular reactor.

Real reactors are neither ideal. Danckwerts' 1953 paper introduced the residence time distribution, a way to characterise how long fluid elements actually spend inside a vessel, measured by injecting a tracer pulse and recording the outlet response. A real tank might show dead zones or bypassing; a real tube shows some axial dispersion. The tanks-in-series and dispersion models fit such data with a single parameter between the CSTR and PFR limits. CSTRs remain popular despite their volume penalty, because perfect mixing makes temperature easy to control and suits liquid-phase reactions where the reactor volume is cheap.

References

1. Levenspiel, O. (1999). Chemical Reaction Engineering (3rd ed.), Chapters 5–6. New York: Wiley.
2. Fogler, H. S. (2016). Elements of Chemical Reaction Engineering (5th ed.), Chapter 2. Boston: Prentice Hall.
3. Danckwerts, P. V. (1953). Continuous flow systems: Distribution of residence times. Chemical Engineering Science, 2(1), 1–13. doi:10.1016/0009-2509(53)80001-1