First-order systems in series and interacting systems
Non-interacting and interacting first-order systems in series: their transfer functions, effective time constants and S-shaped overdamped step responses.
Drafted with Aria, reviewed by the AiCanCode.org team. Spotted an error? Use Give Feedback at the bottom of the page.
Why it matters
Real plants are chains of capacities: tanks in series, trays in a column, a reactor followed by a cooler, a process followed by a sensor lag. Chaining first-order elements creates S-shaped, sluggish responses that look like dead time, and whether the elements interact changes how sluggish the chain is. This is the bridge from simple first-order behaviour to the second-order and dead-time models used for tuning.
Key ideas
Non-interacting systems. Two capacities are non-interacting when the output of the first drives the second but the second does not affect the first. Example: tank 1 discharges freely (over a weir or through a valve into the top of tank 2), so its outflow depends only on h₁. The overall transfer function is simply the product: G(s) = G₁(s)·G₂(s) = K₁K₂ / ((τ₁s + 1)(τ₂s + 1)).
Interacting systems. When tank 1 discharges into the bottom of tank 2 through a resistance, the flow between them is q₁ = (h₁ − h₂)/R₁; the level in tank 2 pushes back on tank 1. The transfer function is no longer a product of the separate first-order lags. For linear resistances, Q₂(s)/Q(s) = 1 / (τ₁τ₂s² + (τ₁ + τ₂ + A₁R₂)s + 1), with τ₁ = A₁R₁ and τ₂ = A₂R₂. The extra term A₁R₂ in the middle coefficient is the signature of interaction.
Consequences.
- Both kinds are second-order and overdamped; the poles are real and negative, so neither can oscillate after a step (ζ ≥ 1 for non-interacting, ζ > 1 for interacting).
- Interaction spreads the two effective time constants further apart and makes the response more sluggish than the same tanks without interaction.
- The step response of two or more lags is S-shaped: zero initial slope, an inflection point, then an exponential approach. The more lags in series, the more the response looks like a dead time followed by a first-order rise, which is why multi-capacity processes are often fitted with a first-order-plus-dead-time (FOPDT) model.
- When one time constant is much larger than the others, it dominates; the small ones behave roughly like extra dead time.
- The sum of the time constants is not a time constant of the chain. It equals the mean delay (first moment of the impulse response) of a non-interacting chain, useful for approximations only.
N equal tanks. N non-interacting stages of time constant τ give 1/(τs + 1)^N. As N grows with Nτ fixed, the response approaches a pure dead time Nτ (plug flow), linking to the tanks-in-series model in reaction engineering.
Second-element step response. For non-interacting lags with τ₁ ≠ τ₂ and a unit step, y(t)/K = 1 − (τ₁e^(−t/τ₁) − τ₂e^(−t/τ₂))/(τ₁ − τ₂).
Formulas
G(s) = K₁·K₂ / ((τ₁·s + 1)·(τ₂·s + 1))
- Non-interacting lags in series; K gains, τ time constants (s or min).
y(t) = K·M·[1 − (τ₁·e^(−t/τ₁) − τ₂·e^(−t/τ₂)) / (τ₁ − τ₂)]
- Step of size M, τ₁ ≠ τ₂, K = K₁K₂.
y(t) = K·M·[1 − (1 + t/τ)·e^(−t/τ)]
- Two equal lags (τ₁ = τ₂ = τ).
Q₂(s)/Q(s) = 1 / (τ₁·τ₂·s² + (τ₁ + τ₂ + A₁·R₂)·s + 1)
- Interacting tanks with linear resistances; τ₁ = A₁R₁, τ₂ = A₂R₂, A area (m²), R resistance (s/m²), Q volumetric flow (m³/s). Multiply by R₂ for H₂(s)/Q(s).
τ_a·τ_b = τ₁·τ₂ and τ_a + τ_b = τ₁ + τ₂ + A₁·R₂
- Effective time constants τ_a, τ_b of the interacting system: the negatives of the reciprocals of the two real poles.
Worked examples
Example 1 (standard): two non-interacting tanks. Tank 1: A₁ = 2 m², R₁ = 0.5 min/m². Tank 2: A₂ = 1 m², R₂ = 0.5 min/m². Tank 1 discharges freely into tank 2. The inlet flow to tank 1 steps up by M = 0.4 m³/min. Find the change in h₂ after 1 min and at steady state.
- τ₁ = A₁R₁ = 1.0 min; τ₂ = A₂R₂ = 0.5 min.
- Gain from inlet flow to h₂: K = 1·R₂ = 0.5 m per (m³/min); final change Δh₂(∞) = K·M = 0.5·0.4 = 0.20 m.
- At t = 1 min: τ₁e^(−1/1) = 0.3679; τ₂e^(−1/0.5) = 0.5·0.1353 = 0.0677.
- Bracket = 1 − (0.3679 − 0.0677)/(1.0 − 0.5) = 1 − 0.6004 = 0.3996.
- Δh₂(1 min) = 0.20·0.3996 = 0.0799 m.
Δh₂(1 min) ≈ 0.080 m; Δh₂(∞) = 0.20 m
Example 2 (GATE level): effect of interaction. Two identical tanks with A = 1 m² and R = 1 min/m² (so τ₁ = τ₂ = 1 min, A₁R₂ = 1 min) are connected (a) non-interacting and (b) interacting. For a unit step in inlet flow, find the effective time constants and the fraction of the final outlet-flow change reached after 1 min.
- (a) Non-interacting: Q₂/Q = 1/(s + 1)². Time constants 1 and 1 min. Fraction at t = 1: 1 − (1 + 1)·e^(−1) = 1 − 0.7358 = 0.264.
- (b) Interacting: denominator τ₁τ₂s² + (τ₁ + τ₂ + A₁R₂)s + 1 = s² + 3s + 1.
- Roots: s = (−3 ± √5)/2 = −0.382 and −2.618 min⁻¹, so τ_a = 2.618 min and τ_b = 0.382 min (check: product 1.000, sum 3.000).
- Unit step: y(t) = 1 − (τ_a·e^(−t/τ_a) − τ_b·e^(−t/τ_b))/(τ_a − τ_b).
- At t = 1: τ_a·e^(−0.382) = 2.618·0.6825 = 1.7868; τ_b·e^(−2.618) = 0.382·0.0729 = 0.0279; y = 1 − (1.7868 − 0.0279)/2.236 = 1 − 0.7866 = 0.213.
(a) 26.4 % and (b) 21.3 % of the final change after 1 min: interaction makes the response slower, because one effective time constant grows to 2.618 min.
Common mistakes
- Adding time constants and calling the sum "the" time constant of the chain.
- Multiplying the individual lags for an interacting system; the A₁R₂ term is then lost.
- Claiming interacting tanks can oscillate: with only resistances and capacities, the poles are always real.
- Using the general two-lag formula with τ₁ = τ₂ (division by zero); use the equal-lag form instead.
- Forgetting that the gain to a level includes R₂, while the gain to the outlet flow is 1.
For GATE CH
Expect: deriving or recognising the interacting-tank transfer function, finding the poles or effective time constants, computing the response of the second tank at a given time, comparing interacting with non-interacting arrangements, and identifying the order of a chain of capacities. Practise the quadratic-root step and the two-lag step response formula.
Quick check
- What is the transfer function of three equal non-interacting lags of τ = 2 min?
- Which is faster after a step: two interacting or two non-interacting tanks with the same A and R?
- Can two non-interacting first-order lags in series give overshoot?
- For interacting tanks with τ₁ = 2, τ₂ = 1 and A₁R₂ = 2 (min), what is the middle coefficient of the denominator?
Answers: 1. 1/(2s + 1)³. 2. The non-interacting pair. 3. No, it is overdamped (or critically damped for equal τ). 4. 2 + 1 + 2 = 5 min.
Interview questions
All Process Instrumentation and Control interview questionsTry answering each one aloud before you open it.
1.Explain what is meant by 'first-order systems in series'.Concept
It means first-order elements arranged so the output of one is the input of the next, such as tanks discharging one into another, or a process followed by a sensor lag. If they are non-interacting, the overall transfer function is the product K₁K₂/((τ₁s + 1)(τ₂s + 1)), a second-order overdamped system. Its step response is S-shaped with zero initial slope, and with many lags in series it looks like a dead time followed by a first-order rise.
2.What is an interacting system in process control?Concept
Two capacities interact when the downstream one affects the upstream one: for example, tank 1 discharges through a resistance into the bottom of tank 2, so the flow between them is (h₁ − h₂)/R₁ and the level in tank 2 pushes back on tank 1. The overall transfer function is then 1/(τ₁τ₂s² + (τ₁ + τ₂ + A₁R₂)s + 1) rather than the product of the separate lags. The system stays overdamped but becomes more sluggish than the same tanks without interaction.
3.Why are first-order systems in series used in process control?Application
First-order systems in series are used in process control to model more complex systems that cannot be accurately represented by a single first-order system. By connecting multiple first-order systems in series, engineers can approximate the behavior of higher-order systems, allowing for more precise control and prediction of system dynamics.
4.What happens if two first-order systems in series have significantly different time constants?Application
If two first-order systems in series have significantly different time constants, the system with the larger time constant will dominate the overall response. The slower system will determine the speed of the overall system's response, as it takes longer to reach its final value compared to the faster system. This can lead to a sluggish overall system response.
5.How can you determine if a system is interacting or non-interacting?Application
Ask whether the flow (or heat or mass transfer) out of the first element depends on the state of the second. If tank 1 discharges freely, for example over a weir or into the top of tank 2, the outflow depends only on h₁ and the system is non-interacting; the transfer function is the product of the individual lags. If the outflow depends on h₁ − h₂, the system is interacting and the denominator contains the extra A₁R₂ term in the coefficient of s.
6.Two non-interacting first-order lags with τ₁ = 2 s and τ₂ = 3 s are in series. What is the overall transfer function, and is there an overall time constant?Numerical
The overall transfer function is K₁K₂/((2s + 1)(3s + 1)) = K₁K₂/(6s² + 5s + 1), a second-order overdamped system with τ = √6 = 2.45 s and ζ = 5/(2√6) = 1.02. There is no single time constant: the sum 2 + 3 = 5 s is only the mean delay of the response, sometimes used for a rough first-order approximation.
7.What are the challenges in controlling interacting systems compared to non-interacting systems?Application
For interacting capacities, one effective time constant grows larger than either individual τ, so the process is more sluggish and allows less aggressive controller tuning. The model also cannot be built by simply multiplying separate element models, so each unit's dynamics must be identified together. In multi-loop plants, interaction between loops (for example top and bottom composition loops in a column) means that one controller's action disturbs the other, which calls for pairing analysis or decoupling.
Finished this topic? Mark it so your progress, study plan and readiness keep up.
Stuck on something here?