It’s All About Depth

Successful flow measurement studies are dependent on accurate depth readings.

It’s All About Depth

A hydrograph charting a nearly 9-month data set in which the relationship between the two meters changed several times as highlighted by the color bands.

In the inflow and infiltration business, depth is the key to flow measurement success.

It is important to recall that no open-channel flowmeter on the face of the earth measures flow. They all measure depth and velocity and then calculate a rate of flow. In the last issue, we discussed the importance of reviewing meter data in a scattergraph format to understand the hydraulic condition in a sewer, and in this article, we will examine how the scattergraph can be used to evaluate the accuracy of the depth measurement. 

The open-channel flowmeters on the market today deploy at least seven velocity technologies including Acoustic Doppler, Gated Acoustic Doppler, Laser Doppler and Time-of-Travel. But there are only two technologies commonly used to measure depth: pressure transducers and ultrasonic sensors. The vast majority of open-channel flowmeters in use in the country today are equipped with pressure transducers, primarily because they are less expensive and are a bit easier to install. In the open-channel flowmeter business today, much of the marketing is based on patented velocity technologies used in the meter and little mention is made of the mostly unpatented depth technologies. 

Catch the drift

There is a wide range of costs in available pressure transducers, but the types typically deployed in open-channel meters share the common characteristic of sensor “drift.” Pressure sensor drift refers to the tendency for the sensor’s reading to gradually change from its true depth. The ultrasonic sensor is based on a precise clock measuring the return time of an acoustic echo, and its readings tend to have a narrower band of precision and do not drift. Figure 1 is an example of depth measurements in an 8-inch sewer and with both ultrasonic and pressure sensor readings. 

In general, an ultrasonic sensor has greater precision than a pressure sensor, and the difference is displayed in the heights of the red and green bars in Figure 1, which shows 2 inches of water in the 8-inch pipe with the pipe marked in 1/2-inch increments. The flow rate in gallons per minute is shown on the right side of the depth markings. If sensors are installed in the month of March, both would properly measure 2 inches of flow at 24 gpm. During the following 4 months, the pressure sensor drifted by 1/4 inch per month. Drift at such a small rate will likely go unnoticed by the operator, but after 4 months of operation, the meter will be reporting 2 inches of depth (24 gpm) as 3 inches of depth (52 gpm). That is an error of 117 percent.

Vivid data

Pressure sensor drift is hard to spot when it occurs in a single meter, but it becomes vivid when meter subtractions are set up to calculate I&I. When one meter flows into the next, I&I is determined by subtraction of the two meters. There are three separate conditions in which the upstream flow is higher (bad), higher than the upstream (good), and equal to the upstream (not good). The last installment in this series discussed the importance of depth-velocity scattergraphs as a key performance indicator in evaluating flow data, and that is the tool to use to diagnose this problem. 

If all the data on the scattergraph lines up on the pipe curve, the meter is working correctly. If there is depth drift, it will show up as a vertical shift in the pattern along the depth axis.

The example in Figure 2 is of a nearly 9-month data set in which the relationship between the two meters changed several times as highlighted by the color bands along the hydrograph. The downstream meter on the left experienced a small amount of drift, but is generally acceptable. The upstream meter experienced a maximum of 3 inches of drift and is responsible for most of the flow imbalance problem. 

Reviewing the scattergraph of all flowmetering should be the “do-not-pass-go” step. This means that no flow data should be considered ready to use until the data have been viewed in the depth-velocity scattergraph. This is a critical quality assurance/quality control step that should not be skipped. 

Check the volume

A comment that is commonly heard among casual flow data users is “close enough is good enough.” This implies that one does not need to take the time and make the effort to collect accurate or even valid flow data. But as we have seen, a relatively small amount of pressure sensor drift can result in dramatically wrong conclusions. In the next installment in this series, we will discuss a second KPI in the I&I business and that is the Q vs. i plot. Q is I&I volume, and i is the rainfall. 

The Q vs. i plot is simply the comparison of rainfall on the bottom axis and the I&I volume on the side axis. For data collected in the same season (e.g., winter or summer), these data points should be linear in nature. The slope of the linear regression is an indicator of I&I severity; the steeper the line, the more severe the I&I. 

Figure 3 is the Q vs. i plot of a small 1-inch pressure sensor drift. The rainfall-dependent I&I volume from the sewershed between these two meters is determined by subtraction. A review of the Q vs. i plot displayed in Figure 3 shows that the relationship is far from the expected linear relationship. The regression line splits the difference between the data points from the two largest storms, but think of the danger if the study captured only one of the two largest storms. The sewer agency could have concluded that this sewershed had little I&I (low slope) if only the March 29 storm was captured or have extremely severe I&I (high slope) if only the March 22 storm was captured. This could have caused the agency to overlook a bad sewershed or spend money trying to rehabilitate a sewershed that was adequate.  

As old-timers in the I&I business (Patrick Stevens is the old-timer), we have learned how to tell the past experiences of a sewer agency by the language in their procurement documents for flowmetering services. If they have been burned by pressure sensor drift in the past, they will specify that the provider must visit the site weekly or biweekly to conduct manual depth measurements. Any difference between the meter depth measurement and the manual measurement must be corrected by adjusting the meter. This has been the only way to spot and control pressure sensor drift. 

Rules of thumb

The old-timers have seen flow data from well over a thousand I&I projects and have developed three rules of thumb about open-channel flowmeters:

Rule 1 - All pressure sensors drift at some time in their life, and the user cannot predict when it occurs or by how much it will drift. 

Rule 2 - When a data set comes from an agency that is not aware of pressure sensor drift and has not tried to control it, much of the data is not usable. The data fall into the 30-40-30 rule of thumb.

30 - Percentage of data with accuracy of 20 percent or better and are useful

40 - Percentage of data with accuracy between 20 and 50 percent and is marginally useful

30 - Percentage of data with accuracy worse than 50 percent and is generally not useful.

Rule 3 – Once a flow data set with pressure sensor drift has been collected, there is no going back to try to correct/recover data.

With the advent of wireless data collection and deployment of ultrasonic depth sensors, drift does not occur and the analyst can spot any discrepancy in meter performance on almost a daily basis by the use of scattergraphs. Many sewer agencies today will specify that flowmeters deploy both depth sensor technologies: ultrasonic depth during open-channel condition and pressure sensors for when the sewer surcharges. 

It is important to recognize that the accuracy of a flow rate calculation is influenced by the depth measurement much more than by the velocity measurement. The velocity sensors on the market tend to deliver similar results that may differ by 10 or perhaps 20 percent. So that is the maximum flow rate error that can be generated by the velocity sensor. But as we have seen in this article, flow rate error that can be generated by the depth sensor can be 100 to 200 percent, especially in shallow flow.


Comments on this site are submitted by users and are not endorsed by nor do they reflect the views or opinions of COLE Publishing, Inc. Comments are moderated before being posted.