Biotechnology · Advanced

Fed-batch fermentation

Written by SignalsLab

Early edition: these lessons are being reviewed.

Content

Summary

Fed-batch fermentation adds nutrient feed to a bioreactor during the run without removing broth, so volume rises while substrate is kept low. This avoids overflow metabolism, substrate inhibition and oxygen limitation, and allows high cell density and product yield. The classic strategy is an exponential feed that holds specific growth rate constant.

Intuition

Imagine feeding a fire: dumping all the wood at once creates smoke and waste (acetate, ethanol, heat), but adding wood at the rate the fire can burn keeps it clean and hot. In a fed-batch, the cells are the fire and the substrate is the wood. Because the feed is the limiting factor, the operator, not the cells, sets the growth rate. Substrate stays near zero, the cells consume it as fast as it arrives, and biomass accumulates. Volume grows, so concentrations are diluted, but total biomass and product mass keep rising until the vessel is full or oxygen transfer limits the culture.

Key equations

dVdt=F
(1)

Volume rises at the feed rate F (no outflow).

d(VX)dt=μVX
(2)

Total biomass mass balance; X is biomass concentration.

d(VS)dt=FSF−μYX/SVX−mVX
(3)

Substrate balance with feed concentration S_F, yield Y_{X/S}, maintenance m.

F(t)=μsetYX/S(1SF−S)...≈(μsetYX/S+m)X0V0SFeμsett
(4)

Exponential feed profile keeping quasi-steady S and constant specific growth rate.

Derivation

  1. Since there is no outflow, total volume obeys dVdt=F.
  2. Biomass mass: d(VX)dt=μVX. Expanding gives VdXdt+XF=μVX, so dXdt=(μ−D)X with D=F/V.
  3. Substrate mass: d(VS)/dt = F·S_F − (μ/Y + m)VX.
  4. Assume quasi-steady state with S small (d(VS)/dt ≈ 0, S ≪ S_F) and μ = μ_set under substrate limitation.
  5. Then F·S_F = (μ_set/Y + m)VX, so F = (μ_set/Y + m)VX/S_F.
  6. Since VX grows as e^{μ_set t}, F(t) = F_0 e^{μ_set t} with F_0 = (μ_set/Y + m)X_0V_0/S_F.

Worked example

Problem

A fed-batch starts with V0 = 2 L, X0 = 10 g/L. Target μ = 0.15 h^-1, Y_X/S = 0.5 g/g, m = 0.02 g/g/h, feed S_F = 400 g/L glucose. Find the initial feed rate and the feed rate after 10 h.

  1. Substrate demand per gram biomass: μ/Y + m = 0.15/0.5 + 0.02 = 0.32 g/g/h.
  2. Biomass at start: X0V0 = 10 × 2 = 20 g.
  3. Glucose demand: 0.32 × 20 = 6.4 g/h.
  4. F0=6.4/400=0.016 L/h.
  5. After 10 h: F = 0.016 × e^{0.15×10} = 0.016 × 4.48 = 0.0717 L/h.

Answer F0 = 0.016 L/h; F(10 h) ≈ 0.072 L/h

Common mistakes

  • Treating concentration X as if it rises with the same formula as in a chemostat; volume dilution matters, so track total mass VX.
  • Forgetting maintenance m, which makes the feed too low and lets μ fall below target.
  • Continuing exponential feed beyond the oxygen transfer or cooling capacity of the reactor.
  • Assuming S is zero exactly; it is small but nonzero, set by Monod kinetics.
  • Confusing fed-batch with continuous culture: no broth is withdrawn, so no steady state exists in volume.

Animation

Animation coming soon

The animated version of this lesson is being prepared for the web. In the Signals AI app, lessons are drawn step by step.

Code

scene.py(72 lines)
import sys
import numpy as np
from manim import *
from blocks import *


def simulate(dt=0.05, T=20.0):
    """Fixed-step Euler fed-batch: Monod growth, constant feed F into volume V."""
    mu_max, Ks, Yxs, S_in, F = 0.4, 0.5, 0.5, 20.0, 0.1
    V, X, S = 1.0, 0.5, 10.0
    ts, Vs, Xs, Ss, mass = [0.0], [V], [X], [S], [X * V]
    n = int(round(T / dt))
    for k in range(n):
        mu = mu_max * S / (Ks + S)
        dX = mu * X - F / V * X
        dS = -mu * X / Yxs + F / V * (S_in - S)
        dV = F
        X += dt * dX
        S = max(S + dt * dS, 0.0)
        V += dt * dV
        ts.append((k + 1) * dt)
        Vs.append(V)
        Xs.append(X)
        Ss.append(S)
        mass.append(X * V)
    return np.array(ts), np.array(Vs), np.array(Xs), np.array(Ss), np.array(mass)


TS, VS, XS, SS, MS = simulate()


class Main(Scene):
    def construct(self):
        # title
        TitleCard(self, "Fed-batch fermentation", "Monod growth with constant feed")
        # plot of computed values
        idx = np.arange(0, len(TS), 4)
        axes = Axes(x_range=[0, 20, 5], y_range=[0, 12, 4], x_length=8.5, y_length=4.2,
                    axis_config={"color": C_DIM, "include_tip": False})
        axes.move_to([-1.2, -0.3, 0])
        xl = Text("time (h)", font_size=24).next_to(axes, DOWN, buff=0.2)
        yl = Text("g/L, L", font_size=24).next_to(axes, UP, buff=0.1).align_to(axes, LEFT)
        self.play(Create(axes), FadeIn(xl), FadeIn(yl))

        def curve(arr, color):
            pts = [axes.c2p(TS[i], arr[i]) for i in idx]
            return VMobject(color=color).set_points_as_corners(pts)

        cX, cS, cV = curve(XS, C_GOOD), curve(SS, C_BAD), curve(VS, C_MAIN)
        keys = VGroup(
            Text("biomass X", font_size=24, color=C_GOOD),
            Text("substrate S", font_size=24, color=C_BAD),
            Text("volume V", font_size=24, color=C_MAIN),
        ).arrange(DOWN, aligned_edge=LEFT, buff=0.2).to_edge(RIGHT, buff=0.5).shift(UP * 0.5)
        self.play(FadeIn(keys))
        self.play(Create(cS), run_time=2)
        self.play(Create(cX), run_time=3)
        self.play(Create(cV), run_time=2)
        eq = MathTex(r"\frac{dV}{dt}=F").scale(0.8).next_to(keys, DOWN, buff=0.5)
        self.play(Write(eq))
        self.wait(1)
        StepCaption(self, "Feeding substrate keeps S low while volume and biomass grow.")
        self.wait(1.5)


if __name__ == "__main__" and "--selftest" in sys.argv:
    assert abs(TS[0]) < 1e-12 and abs(VS[0] - 1.0) < 1e-12
    assert abs(VS[-1] - (1.0 + 0.1 * 20.0)) < 1e-6
    assert np.all(SS >= 0) and np.all(XS > 0)
    assert abs(MS[-1] - XS[-1] * VS[-1]) < 1e-9
    assert MS[-1] > MS[0]
    print("ok")

The Manim scene code that draws this lesson's animation in the Signals AI app (Python, Manim Community).

ccBYNC© 2026 SignalsLab. Lesson text shared as is under CC BY-NC 4.0.