Thermocouple Transient Response Simulator — First-Order Lag
Models a temperature sensor as a first-order lag system and visualizes the indicated temperature over time. Vary the time constant, initial temperature, and fluid temperature to learn the behavior of measurement lag.
Parameters
Initial temperature T₀
°C
Fluid temperature T∞
°C
Time constant τ
s
Elapsed time t
%
"Elapsed time t" is the percentage of 5τ (nearly settled).
Results
—
Current indicated temperature T
—
Response ratio
—
Residual (T∞ − T)
—
90% settling time
Sensor and Fluid Temperature
Color = temperature (blue = cold, red = hot) / the sensor following the fluid temperature
Indicated Temperature over Time T(t)
Horizontal axis = elapsed time t / Vertical axis = indicated temperature T (yellow dot = current time, dashed = fluid temperature T∞)
Theory & Key Formulas
Models a temperature sensor as a first-order lag system with lumped heat capacity. This is the response when a sensor at initial temperature T₀ is plunged into a fluid at temperature T∞.
Indicated temperature over time (exponential response):
It reaches 63.2% at t = τ, 95.0% at t = 3τ, and 99.3% at t = 5τ. The 90% settling time is t = τ ln(10) ≈ 2.303 τ.
What is the Thermocouple Transient Response Simulator
🙋
When you put a thermometer under your arm, it doesn't show the right temperature right away — you have to wait a while. Is a thermocouple the same idea?
🎓
It is exactly the same. Because the sensor itself has heat capacity, it takes time to "catch up" to the surrounding temperature. This is called a first-order lag system. The indicated temperature approaches the final value as the exponential function $T(t) = T_\infty + (T_0-T_\infty)e^{-t/\tau}$. In the simulator above, if you make the "time constant τ" large and press "Animate response," you can see the tracking become slow.
🙋
What exactly does that "time constant τ" represent?
🎓
τ is a measure of the speed of the response. At t = τ, the temperature difference has shrunk by exactly 63.2%. In formula terms, $\tau = mc/(hA)$. The larger the sensor mass m and specific heat c — that is, the larger the "capacity to store heat" — the larger τ and the slower the response. Conversely, the larger the heat transfer coefficient h and surface area A, the smaller τ and the faster the response. That is why a thin, small sensor is faster.
🙋
I see! Looking at the "Residual" card, it never quite reaches zero. Does it never catch up?
🎓
In theory, yes — because it is an exponential function, it approaches asymptotically over infinite time. But in practice, the residual becomes less than 1% at t = 5τ, so it is considered settled. What matters in the field is when the fluid temperature keeps changing. The sensor always follows with a lag of τ, so if you want to accurately measure a rapid temperature change, you need a sensor with a small time constant. For example, measuring the pulsation of combustion gas requires a fast, fine-wire thermocouple.
Frequently Asked Questions
The lumped-capacitance model holds when the temperature inside the sensor can be assumed uniform (the same regardless of position). This is valid when the Biot number Bi = hL/k is about 0.1 or less, and it holds in many practical conditions if the sensor is small and made of a high-conductivity metal. If Bi is large, the internal temperature distribution of the sensor must be considered.
It is determined by a step-response test. The sensor is suddenly plunged into a medium of known temperature, and the indicated temperature over time is recorded. The time it takes for the response ratio to reach 63.2% is the time constant τ. Alternatively, plotting the residual on a semi-log graph gives a straight line, and the reciprocal of its slope is τ.
Putting the sensor in a protection tube increases the heat capacity and adds a thermal resistance between the tube and the element, so the time constant becomes larger and the response slower. While the time constant of a bare wire thermocouple is tens of milliseconds, a sheathed type can be seconds, and with a thick protection tube it can be tens of seconds. It is a trade-off between response speed and mechanical and chemical protection.
In a pure first-order lag system, the indicated temperature approaches the final value monotonically and no overshoot occurs. Overshoot is a phenomenon that appears in second-order and higher systems. The lumped-capacitance temperature sensor treated by this simulator is a first-order lag system, so the indicated temperature always changes monotonically.
Real-World Applications
Process control and temperature measurement: When selecting a temperature sensor for chemical plants, air conditioning, engine testing, and so on, the time constant is one of the most important specifications. Because the responsiveness of a control loop is directly constrained by the sensor's time constant, a fast sensor is essential when fast control is needed.
Sensor calibration and dynamic compensation: If the sensor's time constant is known, the lag of the reading can be numerically compensated to estimate the true temperature. This is called dynamic compensation, and it is used as a technique to effectively speed up the response of a slow sensor through signal processing.
Control engineering education: The first-order lag system is the first basic element learned in control engineering. Because concepts such as the time constant, step response, and settling time can be learned through a familiar temperature sensor, it is treated as an example problem in many textbooks. Charging an RC circuit and water-level control of a tank are also first-order lag systems.
Fire detection and safety devices: In the heat-sensitive element of a sprinkler or the heat sensor of a fire alarm, the time constant determines the response time. Standards specify the operating time for a given rate of temperature rise, which is essentially a response calculation of a first-order lag system.
Common Misconceptions and Cautions
The most common misconception is to assume the time constant τ is the "time at which the response completes." At t = τ, the response is still only 63.2%, with more than a third of the temperature difference remaining. It can be said to be "nearly settled" only around t = 5τ (residual 0.7%). Watch the response ratio card in the simulator and check the values at τ, 3τ, and 5τ. τ is a "measure of speed," not a "completion time."
Next is the mistake of thinking the difference between the initial temperature T₀ and the fluid temperature T∞ affects the "speed" of the response. The response ratio $1 - e^{-t/\tau}$ of a first-order lag system depends on neither the temperature difference nor the value of T₀ — it is determined by the time constant τ alone. Whether the difference is 50 °C or 200 °C, with the same τ the response ratio is the same at the same time. A larger difference makes the absolute residual (°C) look larger, but the "speed" of the response itself does not change. Vary only T∞ in the simulator and confirm that the time history of the response ratio is the same.
Finally, the pitfall of thinking you can use the catalog time constant as-is. The time constant published by the manufacturer is a value for a specific medium, flow velocity, and conditions. The time constant is inversely proportional to the heat transfer coefficient h, and h varies greatly with flow velocity and the type of fluid. The time constant in still air can be tens of times the time constant in flowing water. The time constant under actual operating conditions must be measured under those conditions, or a correction of the heat transfer coefficient must be applied.