Ordinary Differential Equation Models

Introduction

Stoichiometry tells how each reaction changes species amounts, and rate laws tell how quickly each reaction proceeds. Together they form ordinary differential equations (ODEs) that predict concentrations over time in a well-mixed system. An ODE model can explain when an intermediate peaks, why a product lags or how changing feed concentration alters selectivity. Its predictions remain conditional on the proposed steps, parameters and reactor boundary.

Core explanation

For species concentration vector c, stoichiometric matrix N and flux vector v(c, T), the compact batch-model equation is dc/dt = Nv. Each component is a production-minus-consumption balance. In A → I → P with first-order elementary laws v₁ = k₁[A] and v₂ = k₂[I], the equations are d[A]/dt = −k₁[A], d[I]/dt = k₁[A] − k₂[I], and d[P]/dt = k₂[I]. Initial concentrations complete the mathematical problem. An ACS educational study of time-resolved kinetic mechanisms uses matrix kinetics to connect such steps to observed profiles.

The equations describe a particular reactor idealisation. A closed, constant-volume batch model has no feed or removal. A continuously stirred tank adds inlet and outlet terms; a plug-flow reactor often uses residence time or spatial position as a variable. A model for atmospheric chemistry may need sunlight and spatial transport. Do not silently use a batch equation for a flowing system.

Numerical integration estimates trajectories when exact algebra is difficult. At each small time step, the model evaluates current rates and updates species amounts. A simple forward Euler step is c(t + Δt) ≈ c(t) + Nv(c(t))Δt. It is easy to understand but can be inaccurate or even produce negative concentrations if Δt is too large for fast reactions. More reliable solvers adapt step size or handle stiff timescales. Check that predicted concentrations stay physical and conserved element totals remain constant within numerical tolerance.

An ODE model can be fitted to concentration-time data, but a good fit does not prove the mechanism. Different reaction networks can produce similar product curves. Include intermediate and reactant measurements, change initial concentrations or temperature, and test predictions outside the fitting conditions. Parameter uncertainty and detector calibration matter; a curve drawn through noisy points can be visually persuasive but mechanistically weak.

Step-by-step reasoning

1. Write balanced steps and a stoichiometric matrix. 2. Assign justified rate laws and units to each step. 3. Add feed, outflow or transport terms for the chosen reactor boundary. 4. Set initial conditions and integrate with an appropriate time step or solver. 5. Check nonnegative concentrations, conserved totals and independent data.

Visual explanation

Draw A falling with time, I rising to a maximum then falling, and P rising with a delay. Beneath the curves draw A → I → P with fluxes v₁ and v₂. At the instant I reaches its peak, its production and consumption rates are equal, so its slope is zero even while both reactions continue.

Real-world analogy

A reservoir's water level changes according to incoming flow minus outgoing flow. Knowing pipe layout alone does not predict the level; flow laws and a starting level are needed. A chemical ODE uses the same balance logic for each species, but rates depend on composition and temperature and can be reversible.

Real-world example

A dye reacts through a coloured intermediate before a colourless product. Spectra at several wavelengths give time courses for starting dye, intermediate and product. An ODE model predicts the intermediate's rise and fall. If it fits only the final colour loss but not the intermediate peak, the proposed rates or steps may be wrong. The intermediate observation gives a stronger mechanism test.

Why?

Why do initial conditions matter? The same network and rate constants can produce different trajectories from different starting mixtures. Starting with some product in a reversible system can drive backward flux; starting with none cannot. A fitted rate constant should be tested against several initial conditions before being called transferable.

Common misconception

“The ODE solver supplies chemical mechanism” is false; the chemist supplies reactions and rate laws. “A smooth simulated curve proves accuracy” ignores assumptions and data. “A zero slope for an intermediate means reactions stopped” confuses balanced fluxes with zero flux. “Negative numerical concentrations are physically meaningful” usually indicate a solver, step-size or model problem.

Worked example

For A → I → P, let k₁ = 0.2 min⁻¹, k₂ = 0.5 min⁻¹ and initial [A] = 1.0 M, [I] = [P] = 0. At time zero, v₁ = 0.2 M/min and v₂ = 0, so derivatives are (−0.2, +0.2, 0) M/min. A forward Euler step of 0.1 min predicts [A] ≈ 0.98 M, [I] ≈ 0.02 M and [P] ≈ 0. The conserved sum remains 1.0 M. At the new state, v₂ = 0.5 × 0.02 = 0.01 M/min, so P begins to rise. These are small-step approximations, not exact concentrations at long times; repeated stable integration is needed for a full curve.

Quick check

1. What is d[I]/dt for A → I → P with fluxes v₁ and v₂? Answer: d[I]/dt = v₁ − v₂.

Exam focus

Construct ODEs by adding incoming and outgoing stoichiometric fluxes. State initial conditions and reactor assumptions. Perform one Euler update with units, then check conservation and nonnegative predictions. Explain why model fit alone does not prove mechanism.

Advanced insight

Stiff systems contain fast and slow timescales, so explicit fixed-step integration can require impractically tiny steps. Implicit or adaptive solvers can handle stiffness more efficiently. Regardless of solver, a model should be checked for conservation, parameter sensitivity and comparison with independent experimental conditions.

Summary

Reaction-network ODEs combine stoichiometry, rate laws and reactor boundaries to predict species time courses. Numerical solutions are useful only when initial conditions, physical constraints and independent validation are stated.

Practice questions

1. What equation combines stoichiometry and fluxes in a closed, well-mixed batch system? Answer: dc/dt = Nv. 2. Why might an intermediate peak before final product reaches its plateau? Answer: Its formation first exceeds consumption, then its consumption catches up and later exceeds formation. 3. What does an Euler step require besides an ODE? Answer: Current state, chosen time step and parameter values or rate constants. 4. Why should a simulation check atom conservation? Answer: It can expose stoichiometric mistakes or numerical integration errors.