How to Calibrate Flowmeters

Welcome! If you are reading this, you are likely responsible for calibratingthe laboratory flowmeters used to measure fluid discharge rates. This guide explains how to calibrate two types of flowmeters used in the system: a hydraulic flowmeter and a paddlewheel flowmeter. The goal is to determine the relationship between the actual flow rate and the instrument outputs, and to evaluate how accurately these devices measure flow.

In this system, the hydraulic flowmeter measures flow indirectly through the pressure difference across the meter, while the paddlewheel flowmeter produces an electrical signal proportional to fluid velocity. The calibration process compares these readings with the true flow rate measured using a weight-time method with a weighing tank, which serves as the experimental standard.

By following these steps, you will generate calibration curves, determine discharge coefficients, and assess the reliability of each flowmeter.

1. Station F4 apparatus: 4-inch pipe with orifice plate (D = 102.3 mm, d = 50.8 mm, β = 0.497)
2. Mercury-water differential manometer and Validyne CD101 pressure transducer
3. Signet 3-8511-P0 paddlewheel flowmeter + Signet 8511 transmitter (Gain = 6.25 turns)
4. LabVIEW data acquisition software
5. Weighing tank (scale weight ratio 200:1)
6. Stopwatch

Before turning on any flow, take 5 minutes to inspect the apparatus.

1. Verify the discharge valve is fully CLOSED. Never open flow before instruments are zeroed.
2. Check the mercury levels in the U-tube manometer. Both columns should be equal at zero flow. If they differ, slowly open and close the two drain valves ("CAL VALVE" and the second drain) to bleed trapped air from the supply lines.
3. Adjust the central scale between the two manometer columns to read exactly zero.

The Validyne differential pressure transducer must be calibrated against the mercury manometer before collecting flow data. This is done with NO flow in the pipe.

1. On the LabVIEW interfrace, navigate to the transducer calibration panel.
2. With the discharge valve still closed, zero the transducer output using the VFn interface box at the computer.
3. Open the "CAL VALVE" slightly to create a small artificial pressure difference in one manometer leg.
4. In LabVIEW, record the transducer output voltage (V) and read both manometer column heights (cm). Calculate Δh, the difference between the column heights.
5. Repeat at four more settings, increasing CAL VALVE opening each time, until you reach the maximum deflection possible. You should have 5 data points total (including zero).
6. IMPORTANT: The maximum transducer voltage must not exceed 10 V. If it does, stop and contact your supervisor.
7. LabVIEW performs a linear least-squares fit (Δh = slope*V + intercept) automatically and stores the results. Verify the fit looks reasonable.
8. Close the CAL VALVE fully before proceeding.

Our calibration result (for reference): slope = 9.49 cm/V, intercept = -0.13 cm.

You will collect data at 10 flow rates ranging from maximum down to approximately 10% of maximum. At each flow rate, you will record manometer deflection, transducer voltage, paddlewheel voltage, and a weight-time measurement.

Finding Maximum Flow Rate

1. Open the discharge valve slowly while watching the manometer. Note the maximum manometer deflection Δh_max achievable without exceeding the manometer range.
2. At maximum flow, record: manometer readings (both columns), paddlewheel voltage, and transducer voltage from LabVIEW.
3. Take a weight-time measurement: start the stopwatch when the weighing tank begins filling, stop it when 500 lbs have been collected. Record the average time t (seconds). The flow rate is Q = (500 lb / 62.43 lb/ft^3) / (200*t), but LabVIEW computes Q for you.
4. In LabVIEW, trigger the time-averaged transducer voltage acquisition over the prescribed interval (7 seconds). Record all values.

Stepping Down Through 9 More Flow Rates

1. Reduce the valve so the manometer deflection is approximately 0.9*0.9*Δh_max. Wait for the mercury to stabilize before recording.
2. At each set point, repeat: record manometer heights, paddlewheel voltage, LabVIEW transducer voltage, and take a weight-time measurement.
3. Continue stepping down: target deflections are 0.8*0.8, 0.7*0.7, 0.6*0.6, 0.5*0.5, 0.4*0.4, 0.3*0.3, 0.2*0.2, 0.1*0.1 times Δh_max, giving approximately 80%, 64%, 49%, 36%, 25%, 16%, 9%, 4%, 1% of maximum flow.

With data collected, you can now produce calibration curves. The key equation relating Q to manometer deflection Δh for an orifice-plate meter is given above.

Step 4a: Plot Q vs. Δh on Linear Scales

The primary calibration curve on linear axes is given. Every point from the data table maps directly onto this plot. The smooth curve is a power law best fit through the data.

Step 4a: Plot Q vs. Δh on Log-Log Scales

Plotting on logarithmic scales reveals whether the data follows a power law Q = K*(Δh)^m. If the data is linear on a log-log plot, a power law applies. Based on the graph, the data falls on an approximately straight line, confirming a power law relationship. Linear regression on the log-transformed data gives Q = 0.00107 *(Δh)^0.688 and R^2 = 0.99. Theoretically, if Cd is constant, the slope should be exactly m = 0.5. Our result of 0.688 is higher than 0.5, suggesting that Cd is not constant, and it decreases at lower flow rates, which is explained further in the next step.

The Reynolds number is defined using the full pipe diameter D and the upstream pipe velocity V1 described above, where V1 is Q / A_pipe. A pipe = (π/4)(0.1023)^2 = 8.22 x 10^-3 m^2, and v = 8.711 x 10^-7 m^2/s at 26°C (from LabVIEW). LabVIEW computes Re and Cd for each data point. The graph of discharge coefficient Cd vs. Reynolds number Re_D is also plotted.

Cd is not essentially constant across the tested range. It drops from 0.65 at high Re to 0.41 at the lowest Re tested. At high Reynolds numbers, the values cluster near the ISO references of 0.61. At lower Re, viscous effects cause Cd to fall significantly. The data point Re = 290,833, Cd = 0.846 appears to be an outlier, possibly due to unsteady manometer readings during that acquisition. The experimentally measured Cd values are close to the ISO references at high Re but deviate at low Re.

The paddlewheel produces a voltage proportional to fluid velocity. The plot of voltage vs. actual Q from the weight-time measurements on the linear axes. The key results are as follows:

1. Linear fit: V_paddle = 236.5*Q + 0.246 (V in volts, Q in m^3/s)
2. Cutoff flow rate: Q_cutoff = 0.00375 m^3/s (lowest Q at which paddlewheel registered a nonzero reading)
3. Cutoff velocity: V_cutoff = Q / A_pipe = 0.456 m/s = 1.50 ft/s
4. Maximum velocity achieved: V_max = 2.39 m/s = 7.84 ft/s

The Signet specifications state a linear range of 0.3-20 ft/s. Our measured cutoff velocity is 1.50 ft/s, which is above the stated lower limit of 0.3 ft/s. This means the paddlewheel stopped responding before reaching its rated lower limit and did not meet low-end specifications under our conditions. The upper end was not reached, so the device likely meets the high-end spec.

Is Cd Constant?

No, Cd is not constant. Cd varies approximately 0.41 to 0.85 across Re = 53,000 to 290,000. Excluding the apparent outlier, Cd ranges from about 0.52 to 0.72. At high Re, Cd stabilizes close to the ISO value. The theoretical value of Cd = 1 is not achieved in practice because of the energy from flow separation and turbulence.

How Reliable is the Paddlewheel?

The Paddlewheel is more reliable at high flow rates. At high Q, the voltage output is large, stable, and clearly above the zero-cutoff noise flow, making readings easy to interpret. At low flow rates, the paddlewheel abruptly drops to zero when Q falls below approximately 0.00375 m^3/s, giving no useful signal. Near the cutoff, readings are unstable and prone to fluctuations as the wheel intermittently rotates. This makes the paddlewheel unreliable for precise low-flow measurement. For best results, always use the hydraulic (orifice-plate) flowmeter for low-flow conditions, and treat paddlewheel readings near the cutoff as unreliable.