mathematics//dynamical systems//nonlinear dynamics

Nonlinear dynamics is the study of dynamical systems whose rates are not proportional to their states, and of the behaviours only such systems can show: several resting states, oscillations that sustain themselves, sudden changes of regime and chaos. An engineer needs it to know where the linear toolbox stops. Almost all industry works near an operating point, linearizes there and applies linear theory, and that is the right call; the nonlinear machinery pays off when the operating range is wide, or when the interesting thing happens far from equilibrium (a manoeuvring drone, a compressor near its limit, a grid under stress).

Nonlinear dynamics is the study of dynamical systems whose rates are not proportional to their states, and of the behaviours only such systems can show: several resting states, oscillations that sustain themselves, sudden changes of regime and chaos. An engineer needs it to know where the linear toolbox stops. Almost all industry works near an operating point, linearizes there and applies linear theory, and that is the right call; the nonlinear machinery pays off when the operating range is wide, or when the interesting thing happens far from equilibrium (a manoeuvring drone, a compressor near its limit, a grid under stress).

The nonlinearities found in a plant are rarely exotic. An actuator saturates, a valve has a dead zone and sticks, a gearbox has backlash, a bearing has dry friction, a tank empties through an orifice whose outflow grows with the square root of the level. Each of them breaks superposition: doubling the input no longer doubles the response, and the behaviour depends on the amplitude.

Nonlinearity brings several equilibria, limit cycles, bifurcations and chaos, and in chaos the prediction horizon grows only with the logarithm of the measurement precision. None of the four can happen in a linear system, so seeing one on a trend is proof that the linear model is the wrong tool for that question.

Multiple equilibria come first. A linear system has one resting state or a whole line of them; a pendulum has two isolated ones, hanging and inverted, one stable and one not, and which one a disturbance leaves you near decides everything after. Each is studied with its own linearization (equilibrium and stability).

A limit cycle is the oscillation to look for in a control loop: a fixed amplitude reached from any start, the trademark of a sticking valve or an on-off controller. In a linear system the amplitude would depend on how the motion began.

A bifurcation is the change of regime: a parameter crosses a threshold and an equilibrium vanishes or an oscillation is born, as when a centrifugal compressor is throttled into surge. It is the note to read for operating limits.

Chaos is determinism without long-term prediction, the reason a weather forecast lasts days. It is also the rarest of the four in machines: most nonlinear machines are perfectly predictable.

The tools change with the question. The phase portrait shows the whole picture in two dimensions, a Lyapunov function proves stability without solving the equations, and beyond that the honest tool is simulation, run over many initial conditions.