Increased emphasis on the need to improve process performance and reduce variability mean many control loops today are optimized. This is done through digital adaptive control and predictive control algorithms in distributed control systems or programmable logic controllers. These advanced control routines make it easy to linearize control valve performance and get the most out of assets. They do not, however, guarantee accurate and repeatable input values from all the primary measuring instruments. That’s why when measuring flow, one of the easiest and most economical ways to ensure that process improvements are what we expect is to make sure our flow devices have a happy home. Let me explain:
Flow meter manufacturers, in accuracy statements for volumetric flow measurement technologies, assume an ideal, fully developed flow profile at the meter inlet. This ideal flow stream is a symmetric, swirl-free, turbulent flow profile. The actual shape of the velocity profile is determined by the viscosity of the fluid, the pipe wall roughness and the Reynolds number. Flow technologies, such as orifice, vortex, ultrasonic and turbine meters, have the same requirements.
A less-than-ideal velocity profile is one that is distorted in some way. For example, a flow stream may have a fully developed profile, but after passing through a 90º elbow, the stream might change, having two counter-rotating vortices and an asymmetric or distorted velocity profile. This nasty, unpredictable flow profile may take 50 pipe diameters of straight pipe to reconcile. Flow distortion can introduce error into the flow rate measurement.
Besides elbows, other common sources for these disturbances are tees, pipe reducers and expanders, but the worst offender is a modulating control valve.
Even control valve performance can be affected by flow disturbances. The ISA Handbook of Control Valves, 1976 shows a control valve installed after an elbow. The recommended spacing shown in that handbook is 16 pipe diameters after a 90º elbow. The requirement is five pipe diameters after the valve before the next elbow (Figure 1).
This critical spacing ensures a steady inlet pressure and flow shape at the control valve. If these are not stable and consistent, flow control will be erratic and the valve pressure drop will be unpredictable.
The recommended straight pipe run required for a flow meter downstream of a control valve is even greater. Several volumetric flow meters require 30-50 pipe diameters. These meter technologies include: orifice flange union, venture, averaging pitot tube, vortex meter, ultrasonic flow meter and turbine meter.
Differential pressure (orifice) meters derive flow rate by taking the square root of the pressure drop created by the orifice.
A non-ideal flow profile would cause unpredictable, nonlinear differential pressures across the plate. The actual amount of error created can range from -2 to +5%, depending on plate design, tap locations and beta ratio. Since the error is not easy to define and isn’t always repeatable, it is very important to give the flow meter a happy home. Both the accuracy and repeatability will be unpredictable because of the erratic pressure and flow profile.
For other flow technologies, including vortex, turbine and ultrasonic, meters report flow based on measuring fluid velocity. The method each of these meters uses to measure velocity and calculate flow will vary significantly, yet each one requires the same ideal flow and pressure profile for accurate meter results. The following equation shows that the orifice flange union flow is dependent on pressure drop. If the pressure drop is unstable, so is the reported flow value.
Single-point insertion meters (Figure 2), pitot tubes and even thermal mass meters are vulnerable to this issue because they read or measure flow at a specific depth relative to the pipe wall. They reside in a velocity region that represents an average velocity of the flow stream. If the flow profile is skewed, the meter will read falsely high or low depending on the distorted shape of the flow stream.
If it is impossible to provide the recommended 20, 30 or more pipe diameters of straight pipe, flow conditioners can be used. These devices, sometimes called flow straighteners (Figure 3), insert into the pipe upstream of the meter and help to restore the ideal flow profile within a much shorter pipe run.
Another concern related to flow profile is the Reynolds number, which is a non-dimensional number that defines if a stream is laminar or turbulent. At laminar flow, the Reynolds number is 2000 and below, the flow stream is dominated by viscous forces, and the flow profile is parabolic-shaped. It’s as if the profile was laminated or flowing in layers. In the center of the flow stream, fluid velocity is very high compared to the velocity closer to the pipe wall, and the velocity profile changes dramatically as flow increases.
At turbulent flow, the Reynolds number is 4000 and above, and dynamic (inertial) forces dominate the flow profile causing the fluid to mix in a way that the velocities are the same across most of the pipe area. The flow profile “squares up”: The fluid velocity and profile is uniform across a large cross-section of the pipe and doesn’t change much as velocities increase except for the boundary layer very near the pipe wall. As fluid moves from laminar to turbulent flow (which is called transitional flow), it becomes very unstable. Its behavior will jump from laminar to turbulent at random. Most flow meters have poor repeatability if operating in this range.
Many users also assume their meters will maintain accuracy and linearity throughout the entire flow range, which is not necessarily the case. For example, vortex meters have a low-flow cutout for this reason. Below turbulent flow, they do not measure at all. Orifice plate users often have a low flow (cutout) point where they artificially linearize the transmitter output from some low-flow value down to zero.
SOLVING THE PROBLEM
If your process acts unpredictably during certain process flow conditions, these areas may be places to solve the problem.
Is the meter in a “transitional flow regime” (fluid velocity is between laminar and turbulent flow)?
Solution: This problem may not be solved if the flow meter technology has a minimum Reynolds number limitation. Consider using a smaller flow meter that can improve the measurement at lower flow rates. In extreme cases, place a second meter in a small bypass line to measure low flows. This solution is only reasonable if, for example, there are seasonal low flow scenarios such as summer vs. winter steam flow rates, which requires diverting the entire flow stream to the bypass line for a season then back to the larger line when demand increases.
Are the flow rates unusually low and possibly below turbulent flow condition?
Solution: If the flow rates are consistently low for the meter size, consider using a smaller flow meter entirely. This will improve the ability to measure at lower flow rates. If these low flows are intermittent, a “low flow cutout” can be configured in the flow meter electronics or in the control system. The flow rate can be set to zero at a minimum flow rate, or a linear output from the flow meter beginning at a minimum flow value to zero can be imposed.
Are flow measurement problems experienced intermittently?
Solution: Intermittent problems are often related to a distorted flow profile (downstream of an elbow, tee or control valve). If that is the case, consider moving the flow meter to another location having additional straight pipe upstream or move the meter completely upstream of the flow disturbance instead of downstream.
These problems cannot be solved with “self-tuning/auto-tune” control algorithms. However, knowing how the flow meter is intended to be used will provide an understanding of whether or not that meter is capable of providing what is expected.
1. ISA Flow Measurement 1996
2. ISA Handbook of Control Valves 1976
3. Article: “Flow Meter Piping Requirements,” Greg Livelli, Flow Control Magazine, Dec. 2015