Why two reactors sized to the exact same average residence time can still deliver completely different conversion. The average is only half the story — the distributionof individual residence times is the other half, and it isn't optional.
It's a natural shortcut: calculate the average residence time a reaction needs, size a vessel — plug flow reactor (PFR) or continuous stirred-tank reactor (CSTR) — to deliver that average, and assume the job is done. For an ideal PFR, that shortcut happens to work, because every fluid element really does spend the same amount of time inside. For a CSTR, it doesn't, because the vessel's defining feature — instantaneous, perfect internal mixing — means individual fluid elements experience wildly different actual residence times even while the average across all of them comes out identical. Matching the average is necessary. It is not sufficient.
The average residence time, τ = V/Q (reactor volume divided by volumetric flow rate), is a single number — the mean time a fluid element spends inside the vessel. The residence time distribution is a completely different, richer piece of information: it describes the actual spread of individual residence times around that mean — how many fluid elements exit early, how many linger, and how many exit right around the average. Two reactors can have identical τ and wildly different RTDs. That difference is exactly what separates an ideal PFR from an ideal CSTR, and it is the RTD — not τ alone — that ultimately governs conversion.
In a CSTR, some fluid elements exit shortly after entering, having barely reacted at all, while others circulate far longer than the average before finally leaving. It's tempting to assume these two effects cancel out — the "extra" conversion from long-staying elements makes up for the shortfall from early-exiting ones, so the average conversion comes out the same as a PFR's. For a reaction whose rate depends strongly on concentration — most reactions above zero order, and especially second-order and autocatalytic kinetics — it doesn't cancel out. The fluid that exits early is under-reacted at the higher concentration it entered with, and there is a hard ceiling on how much extra conversion the long-staying fluid can contribute, because conversion asymptotically approaches completion rather than continuing to climb linearly with time. Net result: the CSTR's wide residence time distribution generally yields lower overall conversion than a PFR at the same average τ. This is a real, well-established consequence of reactor design, not an approximation error — it is precisely why a residence time distribution, not just its mean, is treated as fundamental reactor-engineering data.
False, and it's a genuinely common design mistake. The average residence time, τ = V/Q, can absolutely be made identical between a PFR and a CSTR by sizing the vessels appropriately — that part of the calculation is correct. What that calculation leaves out entirely is the residence time distribution: the actual spread of individual fluid elements' time inside the reactor around that average. An ideal PFR has essentially no spread — every element reacts for exactly τ. An ideal CSTR has a genuinely wide spread — some elements exit almost immediately, some circulate far longer than τ before leaving. For many reaction kinetics, especially higher-order reactions, that distribution difference produces real, meaningfully different conversion between the two reactor types even when their average residence times are identical. Reactor performance depends on the actual residence time distribution characteristic of the reactor type — never just on the average value.
Explains why a plug flow reactor (PFR) and a continuous stirred-tank reactor (CSTR) sized to the exact same average residence time do not necessarily achieve the same conversion — because the actual residence time distribution (RTD), not just its average, is fundamentally different between the two ideal reactor types. Illustrated with a non-mixing plug moving through a PFR, individual fluid-element tracer paths inside a well-mixed CSTR, and a conversion-vs-residence-time comparison for a representative higher-order reaction.
Reactor sizing calculations often start and end with the average residence time, τ = V/Q, because it is the number that shows up directly in a mass balance. It's an easy — and wrong — next step to assume that matching τ between two reactor types means matching performance. The average residence time says nothing about how that time is actually distributed across individual fluid elements, and for a CSTR, that distribution is not a minor detail — it is the defining consequence of perfect internal mixing.
An ideal PFR assumes no mixing in the direction of flow: fluid moves through as a series of thin, non-overlapping "plugs," so every element that enters at the same instant exits at the same instant, after exactly τ. The RTD of an ideal PFR is a spike (a Dirac delta function) at t = τ — essentially zero spread.
An ideal CSTR assumes the opposite: the entire vessel contents are perfectly and instantaneously mixed, so composition is uniform everywhere inside at every moment. Because of that uniform mixing, a freshly entered fluid element has some finite probability of exiting almost immediately (if it happens to reach the outlet quickly) and some finite probability of remaining far longer than τ (if it keeps circulating within the vessel before finally leaving). The RTD of an ideal CSTR is an exponential decay: E(t) = (1/τ)e^(-t/τ). Its mean is exactly τ — identical to the PFR's — but its spread (variance) is not remotely comparable: the CSTR's RTD has substantial probability mass at both very short and very long times, while the PFR's has none.
For a zero-order reaction, rate is independent of concentration, so a wide RTD doesn't change overall conversion much — the CSTR and PFR track closely. For most real reactions, though, rate depends on concentration (first-order, second-order, and higher), and conversion is a nonlinear, concave function of time — it climbs quickly at first and flattens as it approaches completion. Averaging a nonlinear, concave function over a wide spread of residence times (the CSTR case) mathematically yields a lower expected conversion than evaluating that same function at a single fixed time (the PFR case) — a direct consequence of Jensen's inequality. Physically: early-exiting fluid elements in a CSTR are under-reacted, and the extra time given to long-staying elements cannot fully make up the shortfall, because conversion has diminishing returns with time. This is why, for the same average residence time and the same reaction, an ideal PFR virtually always achieves conversion greater than or equal to an ideal CSTR — equal only in the special case of a zero-order reaction.
Yes, in one specific case: a zero-order reaction, where the reaction rate does not depend on reactant concentration at all. Since rate is constant regardless of how long any individual element has been reacting, the spread in residence times inside the CSTR doesn't change the outcome. For any reaction where rate genuinely depends on concentration — the large majority of real reactions — the PFR achieves equal or higher conversion than the CSTR at matched average residence time.
No — that is the idealization. Real tubular reactors have some axial dispersion (a bit of back-mixing along the flow direction) and a real, non-flat velocity profile across the pipe cross-section, both of which introduce some spread around the average residence time. The ideal PFR with zero RTD spread is a useful limiting case; real "plug flow" reactors are engineered (high length-to-diameter ratio, turbulent flow) to approximate that limit closely, not achieve it exactly.
It falls directly out of the perfect-mixing assumption combined with a mass balance on an inert tracer. Because the vessel is uniformly mixed at every instant, the concentration of a tracer pulse injected at the inlet decays exponentially inside the vessel and in the outlet stream over time, which is exactly the exponential distribution E(t) = (1/τ)e^(-t/τ) — mean τ, but with a long tail extending well beyond τ.
Conversion per unit volume isn't the only design consideration. CSTRs are far better at handling highly exothermic reactions (the well-mixed contents make temperature control much easier), are simpler to operate at steady state, handle solids and slurries more readily, and are often the right choice when reaction kinetics are complex or when multiple CSTRs in series are used specifically to narrow the effective residence time distribution while retaining those handling advantages.
A series of N equal-volume CSTRs in a row has a combined average residence time equal to a single CSTR of the same total volume, but a substantially narrower overall residence time distribution — as N increases, the RTD of CSTRs-in-series approaches the sharp, near-zero-spread RTD of a PFR. This is exactly why multiple smaller CSTRs in series are a standard technique for getting CSTR-style temperature/mixing control while recovering much of a PFR's conversion advantage.
Try our Chemical & Process Engineering Studio
More calculators, simulators, and guides for this discipline.