The Canary™ - Theory of Operation

 

There are three main steps in the process of The Canary™ flow measurement.
Step One: Velocity in the pipe is measured.
Step Two: Bulk velocity is calculated based on centerline velocity and Reynolds number.
Step Three: Volumetric flow is calculated based on bulk velocity, pipe diameter, pressure and temperature. 

 

 

 

Velocity Measurement

Figure 1 - Configuration - No particles detected
Single point velocity measurements are performed measuring the time between recorded and filtered pulses. The laser source(1) located inside the flow computer emits laser light in the red region that is guided via a single mode optical fiber(2) into The Canary™ sensor head. Transmitting optics(3) inside the sensing head form two parallel sheets of light in the center of the orifice, side by side, one millimeter apart. The active area for particle excitation is located in the center of the orifice and is approximately 0.5mm x 1.0 mm with a sheet thickness of about 30µm.  Sheets are formed perpendicular to the measured flow.  Forward excitation beams are blocked by the beam blocker(7) to prevent strong light from entering the sensitive receiving optics.  Light scattered from particles passing through the active area is collected by receiving optics(4) and guided into the photo detectors(6) by receiving fibers(5).


Figure 2 - Particle detected passing through beam 1
When a particle passes through the active area of first sheet, light scatters from it. This light is scattered in all directions. The forward part of the scattered light is collected by the receiving optics, coupled into to the first receiving fiber, and then guided into the meter’s first detector where it is detected and converted into an electrical pulse.



Figure 3 – Particle detected passing though beam 2
The identical event occurs when the particle passes through the second sheet, coupling the scattered light into a second receiving fiber, and then guiding it into the second sensor in the flow computer where it is detected and converted into an electrical pulse.


The time between the two detected pulses is measured. The velocity is calculated based on the static, known distance between the laser sheets and the measured time between the resulting pulses.

                                               (1)    v = D / t 

v        [m/s]           Velocity (m/s) of media at measuring point
D       [m]              Distance (m) between centers of laser beams (0.001m)
∆t      [s]               t1-t2 measured time (s) between logged electronic pulses


 
Optical System Description

The Canary™ optical flow meter relies upon a system of lenses, mirrors, and fibers in order to measure velocity and flowrate.  The optical system consists of two major separate systems: transmitting optics and receiving optics.

The transmitting optics begin inside the flow computer. Located inside the flow computer are two excitation lasers that emit 660 nm wavelength light. Light is directly coupled into two single mode fibers.  The fiber guides the excitation laser light into The Canary™ sensing head. Inside the sensor head, light passes through the transmitting optics, through the window, and forms two dense sheets of light one millimeter apart in the center of the orifice.

At the end of the probe is the reflecting mirror. Direct laser light is blocked by the beam blocker placed behind the reflecting mirror in the center to prevent the light from protruding into the receiving channel. The beam blocker is black in color in order to absorb as much light from the laser as possible and reflect as little back as possible.  Scattered laser light from particle excitation is reflected back through the window into the receiving optics.

The receiving channel optics are focused on the same plane as the plane where the excitation lasers are being focused.  When a particle travels through the first laser sheet, the light scatters in all directions from the particle surface. Some of the scattered light, which now can be considered as being emitted from a single point that is on receiving optics focal plane, is focused into the corresponding receiving fiber and guided into the photo detector. The same happens when the particle is passing through the second beam. The two optical signals are coupled into different optical fibers and guided back into the flow computer as separate pulses.

L2F probe


Real Time Signal Processing Parameters

As the light signals from the receiving channels are converted into electrical signals by the photo detectors they are processed by The Canary™ flow computer.  This process creates a signal that can be used for velocity measurements. These parameters are important for setting the conditions by which the raw signals are processed and signal diagnostics are performed. Many of these parameters are updated with each data increment.

The active area refers to the two sheets of light perpendicular to the flow where excitation lasers are focused as well as receiving optics are focused. The Canary’s™ active area is positioned in the center of probe’s flow tube orifice.

The active area has the shape of a rectangular sheet and is approximately 1mm wide x 0.5mm tall and 0.03mm thick. There are two such active areas with a separation of approximately 1mm positioned perpendicularly to the flow.  The areas are designed so that a particle passing through the first active area has high probability of passing through the second active area as well.

Particles passing through the active area cause laser light to scatter, which is then collected by the receiving optics and guided by fiber optics cable into the photo detectors. Only particles passing through the active area can be registered by The Canary™ system as a valid particle pulse and subsequently used in velocity calculations. The system does not detect particles that are not passing through the active area.

There are two properties that can affect the signal strength of the scattered light from a particle that passes through active area. The first property is the size of the particle itself. If the particle is large, more excitation laser light will hit the particle and a greater amount of light will be scattered resulting in a more powerful signal pulse. A smaller particle will create a less powerful signal pulse.  The second property is the power of the excitation laser. Increased power to the laser causes more light to scatter from the same particle. With more powerful laser settings, smaller particle sizes can be detected.

Offset level is an electrical signal representation of the light level permanently present on the receiving channel detectors. There is always an optical background light level being coupled from the transmitting channels into the receiving channels while the lasers are excited. 

The offset levels are directly related to the optical background and are used in separating actual useful signals from random noise and background light present in the receiving channels. This background level does not maintain a perfectly constant value over time. In order to separate useful signals from background noise, the flow computer constantly measures the background level using algorithms to adjust the level automatically every 1ms.

There is a separate offset level for each receiving channel, offset 1 and offset 2, as each receiving channel has different amounts of the background light and thus requires independent compensation.

This background level can also be used as an indicator of probe fouling. If the transmitting optics are fouled, they scatter more light which results in higher background levels.  Background can also be used as detection of dynamic changes in particle types and sizes. For example more liquid and/or larger particles with high concentrations cause offset levels to change to a higher level.

The following image shows a typical signal from the receiving channel.  The automatic offset level is indicated in red dashed line.


Because the offset values do not perfectly approximate the background levels, some of the small noise peaks from the background light can rise above the offset level and resemble a passing particle to the flow computer. This could cause the unit to saturate and read zero velocity since the peaks would not truly represent passing particles but system noise.

For additional protection from the effects of background levels and noise, the threshold parameter is introduced. The threshold parameter establishes a qualification level that the peak of the received pulse must be above to be interpreted as a valid passing particle signal. Any pulses that are registered with a peak signal that are under the threshold level are discarded. 

However the presence of the threshold parameter also causes the electronics to discard real signals. If a particle that passes through does not produce a large enough signal, it does not get registered as a valid particle and therefore is not used for velocity calculations.


The figure above shows a typical signal on the receiving channel during processing.
Offset level is indicated by a black dotted line. Minimum threshold level is indicated by a red dotted line. Minimum threshold is the lowest level of threshold that can be set without noise affecting the signal. This threshold level setting makes the system sensitive to the smallest particles. 

If the threshold level is set to the system minimum, the flow computer will be very sensitive to 0.5-2µm particles. In the configuration above, the system will discard pulse number 1 and it will count pulses 2,3,4,5,6 as valid particles. If the total number of particles counted within a one second period is larger than 100, then the system will automatically increase the threshold in order to decrease the amount of pulses for calculations. In the configuration above, if the threshold level is increased to 13mV pulses 1,2,3, and 6 will be discarded and only pulses 4 and 5 will be used for velocity calculations. 

Threshold levels are handled individually for each receiving channel and are automatically adjusted every second. Algorithms adjusting the threshold levels are finding the balance that ensures that enough particles are being read to establish an accurate velocity reading but not too many particles are being read that would saturate the receiving channels within the flow computer.

Threshold level settings directly affect the number of valid particles registered every period that are used for velocity calculations. Threshold level can be automatically adjusted based on average number of particles during a period. If the average number of particles registered during this period is larger than 100, then the threshold level is multiplied automatically by a factor. If the number of particles is less then 15 during this period then threshold level is divided by a factor.  Multipliers are user adjustable.  Multiplier threshold factor is user selectable between 1-2. Adjustment and proper selection depends on dynamics of the stream where system is installed.

Bin1 and Bin2 are the number of particles that have passed through beam 1 and beam 2 respectively during the sampling period.

Whenever a particle scatters the laser light, that scattered light is received by the receiving optics, coupled into the appropriate receiving optical fiber and guided to the flow computer. Once it arrives at the flow computer, a highly sensitive photo detector measures the light pulse and converts it into an electrical signal. 

A typical pulse signal on the receiving channels is shown below. The offset and threshold levels have been subtracted from the signal.  The pulses remaining are counted as a valid particle signals. The number of valid pulses registered by the flow computer are logged and stored in the “in bins”.  Each set of receiving optics has its own photo detector, which places the converted pulses into the appropriate bin.

There are two bins: bin 1 for the first set of receiving optics – channel 1, which corresponds to the first laser, and bin 2 for the second set of receiving optics – channel 2, which corresponds to the second laser. Every period the bins are emptied and re-evaluated, creating a whole new set of data points.


Once the collection period for the bins has passed, the flow computer must compare the two bins and determine how many correlations exist between the two.

The flow computer is trying to determine how many particles went through both lasers and pair the pulses from each bin. The software in the flow computer compares every pulse in bin 1 to every pulse in bin 2. The computer then calculates the velocities from every one of these valid pairings and creates a histogram out of the results.  The mean of this histogram is then displayed as the measured velocity for that period and the number of calculations used to create this value for velocity is displayed as “InBin”.


The mean velocity is then recorded as raw velocity. A new set of bin counts is measured every period, resulting in a new raw velocity measurement every period.

Post processing filters are designed to smooth the raw velocity readings and filter out incorrect erratic velocity readings. There are three major parameters to set these filters.

At times, possibly due to turbulence in the flow or background readings penetrating into the bin counts, the raw velocity is inaccurate.  These inaccurate velocities can then be filtered out to provide a more accurate, stable velocity reading.

When the raw velocity is measured it is put through various filters that compare it with the previous raw measurements. By these comparisons, the software determines if the flow rate is changing or if the measurement is faulty. If the raw velocity fails to pass through the filters, the raw velocity is intermittently unaltered. However, if too many consecutive readings do not pass through the filter, the raw velocity quickly deteriorates towards a value of zero. 

These parameters allow the user to tweak The Canary™ velocity in order to optimize performance as situational conditions vary.

Bulk Velocity Calculation

The Canary™ measures velocity in a position dependent upon how the probe’s active area is inserted into the flow. In most of the cases the probe’s active area is positioned at a point ¼ of the pipe radius distance from the inner pipe wall.  In other words, the raw velocity measurement is performed at ¼ pipe radius.

For accurate volumetric flowrate calculations, it is necessary to determine the bulk velocity inside the pipe. Bulk velocity is the average velocity inside the pipe across the profile. The velocity profile inside the pipe is a function of the Reynolds number. In order to determine the appropriate correction factor, the ratio between bulk and centerline velocities, it is necessary to determine the Reynolds number. The Reynolds number is a function of pipe diameter, velocity, pressure, temperature and gas viscosity.

The Canary™ calculates every second the Reynolds number, determines the correction factor, and calculates the bulk velocity of the flow in the pipe. The Reynolds number correction table is the primary correction table for The Canary™ . Each flow meter configuration has its own Reynolds number correction table that has been uploaded into the flow computer.

Bulk Velocity Calculations – Reynolds number correction tables
The Canary™ firmware uses the filtered velocity measurement and the Reynolds number to determine the average velocity inside the pipe. Flow profile inside the pipe changes based on Reynolds number as shown below.


When measuring flow, The Canary™ measures velocity, calculates the Reynolds number, determines the correction factor, and then calculates the bulk velocity of the flow.  The Canary™ correction table is based on the following ratio: centerline velocity / bulk velocity = f (Re).  In order to obtain the correct Re number calculation, temperature and pressure parameters must be known.

The graph below is a sample of a calibration table used with a probe mounted on a 10” flow line at the ¼ radius position. It shows the correction factor as a function of the Reynolds number.

Flowrate Calculation
After determining the average flow velocity, the following equation is used to calculate the bulk flowrate:

 

(2)    B = A x v =  (πr2) x v

B       [m3/s] Bulk Flow
A       [m2]            Cross section area of pipe
v       [m/s]           Bulk (average) velocity of the flow which is measured
R       [m]              Radius of pipe where flow is measured


Flow Measurement Range and Accuracy – CEESI Report

Detailed testing of the insertion probe was conducted at CEESI in 2005. In these tests, the flow meter matched the reference within ±2.5% over a 1.0-100 m/s range and within ±7% over the 0.1-1.0m/s range. The flow meter observed increased flow instability within the 0.1 to 1.0m/s range.


Probe Velocity Accuracy and Offset Factor

Every probe manufactured for The Canary™ is tested to determine the measuring properties of each article.  The velocity uncertainty from beam spacing is controlled with accuracy better than 1 μm and the distance between the laser sheets is set at 1.0 mm.  Even through tight manufacturing tolerances, this spacing between the sheets has a small variance creating an offset as shown in blue and varies from probe to probe.  This variance is recorded for every probe by serial number and the resulting correction factor is identified and created in order to set the deviation as shown in purple.  This offset remains fixed over the life of the probe.  The Canary™ does not measure magnitude and therefore has no calibration requirements in order to maintain accuracy and repeatability.  This is a key benefit of this technology.  The probe maintains its accuracy over the life of the product.

 


System Specifications


Pipe Size Installation Range
4” to 30”

Measurement Range
0.1m/s – 150 m/s

Measurement Accuracy
+/- 5% (0.1 m/s - 1m/s)
+/- 2.5% (1 m/s - 100m/s)
+/- 5% (100 m/s - 150m/s)

Measurement Repeatability
+/- 1% (0.1 m/s – 150 m/s)

Turndown Ratio
1:1500

Ambient Temperature
Powerup: -20ºC to +50ºC
Operating: -40ºC to +50ºC

Process Temperature
-40ºC to +100ºC

Maximum Operating Pressure
150 psig

Wetted Materials
Meter Body: 316L Stainless Steel
Optical Windows:  Borosilicate Glass

Additional Integrated Sensors
Temperature and Pressure Sensors

Analog Output Signal
Frequency/Pulse
Current Loop (4 - 20 mA)

Digital Output Signal
WITS (Level 0)
Wireless WITS (Level 0)
TCP (Modbus)
RS-232 (Proprietary Protocol)
RS-485 (Modbus)

System Diagnostics
Low Particle Signal Level Alarm

Datalogging Capabilities
2GB Onboard Storage
1, 3, 5, & 10 Second Selectable Data Rate
FTP Data Retrieval via External Ethernet Port