REACTION ENGINEERING · RESIDENCE
Ten minutes
is not ten minutes.
Residence time is volume over flow: a 100 litre tank fed at 10 litres a minute holds each litre for ten minutes on average. Average is the word doing the work. In a stirred tank the outlet draws from the mixed whole, so some fluid short-circuits in seconds and some sits in a dead corner for an hour. Inject a pulse of dye and watch when it actually comes out.
Fill the tank
The vessel is schematic. Level follows volume on a log scale and the dashes move faster with flow; neither is to scale. Switch the vessel to a tube in the next lab.
VOLUME AND FLOW
Make τ an hour without touching the volume. Then make it six seconds.
100 L fed at 10 L/min: τ = V / Q = 10.0 min. In a plug-flow tube every litre is inside for 10.0 min. In a stirred tank 10.0 min is only the average: 63.2 percent of the fluid is out before then, the median litre leaves at 6.9 min, and 1.8% is still inside after 40.0 min.
Assumptions and limits
τ = V / Q uses the entering volumetric flow; for gases the flow changes with conversion and the mean residence time drifts from τ. Fogler gives τ as equal to the mean residence time in ideal and non-ideal reactors; the slides add the caveats of no density change and no backmixing. The 63.2 percent and the 0.69 τ median are for the ideal stirred tank only. In a real tank dead zones hold fluid that never joins the flow, so the tracer mean comes out below V / Q (MIT 1.85, lecture 4).
- Mean: τ in bothThe ideal stirred tank and the ideal tube share one mean exit time, V / Q. The average hides everything that follows.Fogler, lecture 31
- Median: 0.69 τ against τHalf the fluid in a stirred tank has left by 0.69 τ. In a tube nothing has left until τ, then all of it goes.From E(t) = (1/τ) e^(−t/τ) and E(t) = δ(t − τ)
- First measurement: the curveInject a dye at the inlet, time it at the outlet. Short-circuiting, recirculation and dead zones all show up in the shape before anything else does.MIT 1.85, lecture 4
WHEN DOES IT LEAVE?
The exit-age curve
Inject a dye pulse at the inlet and watch the outlet. In a stirred tank the dye is at the outlet at full strength almost immediately, then fades along a long tail.
E(t) is the exit-age distribution: E(t) dt is the fraction of the fluid leaving between t and t + dt, and the area under the whole curve is 1. The vertical axis is E(t) × τ so the stirred-tank curve starts at 1. The plug-flow spike is drawn with a small width; the ideal one has none.
STIRRED TANK OR TUBE
Set the time to τ itself and read the fraction. Then inject the dye and watch the trace draw.
Stirred tank. The exit draws from the mixed whole, so E(t) = (1/τ) e^(−t/τ): the first fluid leaves at once and the tail runs for hours. By t = 15.0 min (1.50 τ) 77.7% of the pulse has left. Before τ itself: 63.2%, which is 1 − 1/e. The median litre leaves at 0.69 τ = 6.9 min; the mean is still τ = 10.0 min.
Assumptions and limits
Both curves are the ideal cases: perfect mixing in the stirred tank, no mixing along the tube (Nauman 2008). Real vessels sit between them and beyond them: short-circuiting puts dye at the outlet early, recirculation stretches the tail, and dead zones hold fluid out of the flow altogether, which pulls the measured mean below V / Q (MIT 1.85, lecture 4). The dye animation runs at the same screen speed whatever τ is; the axis labels carry the minutes. 63.2 percent is 1 − 1/e and belongs to the ideal stirred tank only.
WHAT THE CURVE SAYS
A continuous reactor is not a fast batch.
Measure the curve before you trust the clock.
τ = V / Q is the mean residence time, and it is the same number for an ideal stirred tank, an ideal tube, and a real vessel with the same volume and flow (Fogler). It is also almost the only thing the three have in common. In the stirred tank the outlet draws from the mixed whole: 63.2 percent of the fluid is out before τ, half of it before 0.69 τ, and nearly 2 percent is still inside after 4τ. The tube sends everything out together, at τ exactly.
A reaction does not see the average. It sees the time each parcel actually spent inside. The parcels that short-circuit come out under-reacted; the ones that linger come out over-reacted, and in a series reaction that means the product you wanted has already turned into the next thing. The same chemistry, the same τ, a different vessel: a different yield.
So the first measurement on a new pilot is the curve, not the conversion. Inject a dye at the inlet, time it at the outlet. Short-circuiting shows up as an early spike, recirculation as a long tail, dead zones as a mean that lands below V / Q (MIT 1.85). Ten minutes is the average. The curve is the truth.
FROM THE FLASK TO THE PILOT
Taking a reaction
beyond the lab?
Jon and I are building RARE Labs toward an autonomous pilot plant builder with AI and robotics. Join the waitlist and tell us what you are trying to scale.
FOLLOW THE EVIDENCE
Sources and model limits
- Fogler, Elements of Chemical Reaction EngineeringLecture 31: residence time distributions ↗
τ defined as V over the entering volumetric flow υ₀, equal to the mean residence time t_m in ideal and non-ideal reactors; the slides' caveats of no density change and no backmixing.
- Residence time distribution (Nauman 2008)Residence time distribution ↗
Ideal CSTR exit ages are exponential, E(t) = (1/τ) e^(−t/τ); ideal plug flow is a Dirac delta delayed by τ. Cites Nauman, Residence Time Theory, Ind. Eng. Chem. Res. 2008.
- MIT OpenCourseWare 1.85, Spring 2006Lecture 4: reactor tanks ↗
The limits of t_R = V / Q for real systems: short circuiting, dead zones and recirculation; the tracer mean equals V / Q only when there are no exclusion zones.