calibration

Hi,

For my WaterCAD model, the client gave us a field flow test result. It seemed that they used a one flow hydrant. I am trying to use this data to calibrate the model, but I am not entirety sure how to do that with only one hydrant. To check for static pressures, I am just running a hydraulic only, steady state scenario and comparing the field and modelled hydrant static pressure? Is that the correct way to do that? The pressure are off by 10 psi which is way to much. So not sure what I am mising.

Now moving on to calibrating for residual pressure when the hydrant is flowed, don't I need a two hydrant test for that?

The main goal for the project is to try and suggest where to put PRVs in the system  to control pressure. The distribution system is very hilly so the pressures are exceeding 100 psi in some areas.

Thanks

  • Hi M, that is a lot of different questions!

    To calibrate pipe head loss characteristics roughness with hydrant tests, you will generally need an upstream reference pressure gauge, a downstream reference pressure gauge on a non-flowing hydrant, and a test flow meter.  Some networks already have a permanently installed and monitored upstream pressure gauge so don't need to use one in the field, or the upstream head source is a tank with known water level RL.

    Hydrant tests don't need to be done to calibrate static pressures.  Normal static pressure readings without any flow can do that.  Unfortunately the source of static pressure errors is rather long in my experience, and is situational but have typically encountered to be issues with either field instrumentation or the network being operated in a way that it wasn't designed for causing it to provide pressures in the field differently to what it should be doing if the network boundary conditions had been set to what they were designed to be set at.

    Possible error sources for which I have encountered through verification processes include:

    • Damaged field pressure gauges where the field team is not conducting a morning calibration verification before doing field tests for the day.  Mechanical Bourdon tube gauges are meant to be carefully looked after and carried around in padded carrying cases, but I see too many contractors just throw them into tool boxes etc.  I had 2 years with 100s of hydrant field tests wrecked by a pressure gauge that was "calibrated" every 12 months but was damaged and was clear to see once I went to site to see what the heck was going on and put the gauge up against a reference gauge to see it was 120 kPa out.
    • Poor quality As-Constructed information for Tank RLs to cause the wrong calculated supply HGL in the model.  Generally you want tank elevations to have been verified by a Surveyor before calibrating.
    • Poor quality Contours / Digital Elevation Models to cause the pressure node in the model to have the wrong Elevation. I generally verify my Client's GIS sources are appropriate and accurate before attempting verification / calibration.   Hilly areas can notoriously have elevation data that is an coarse interpolation rather than real elevations.
    • Wrong PRV or Variable Speed Pump outlet pressure setting in the field for zones supplied from PRV or Pump.
    • Zone Boundary Valve breaches.   Encountered dozens of Zones where Boundary Valves that were supposedly shut to that had been found open or partially open, causing a breach that lowered the pressure in the high pressure zone, and increased the pressure in the low pressure zone.
    • Internal Zone Valve closures:  Valve surveys inside Pressure Zones / District Metered Areas found 100s of valves left shut by maintenance contractors after mains repairs that they had not reopened, and were disguised by alternate flow pathways on ring main systems.  In some cases this cause significant background head losses though in some parts of the day when it was a key distributor that been blocked off.
    • Blocked hydraulic lines from the water main to permanently installed pressure gauges.  They would get a "reading" but the tell-tale is that they would give an abnormally smooth, but virtually never changing value.  These lines need to be regularly flushed of any grit or muck but most maintenance programs don't do this.

    Now in most cases though, verifying whether the field data is "lying" to you is usually simple and doesn't need models.  In a system with very little demand Pressure = HGL Zone @ Pressure Source - Elevation of the Pressure Gauge.  So there are some strategies to check this:

    • In the field go to a hydrant or pressure tapping point near the outlet of the pressure source (Tank, Pump or PRV) where you are confident the Elevation RL is known. 
    • Take two calibrated pressure gauges with you, verify they give the same reading as each other to check for no Pressure Gauge error.  Use the right pressure gauge.  See too many testers use gauges that are meant for hydrostatic max pressure testing of pipes rather than field calibration.  You are generally needing a gauge with a Full Scale Range of no more than ~120 psi, a minimum 4'' dial, and a calibrated error of no more than +/- 1.5% of Full Scale Range.
    • Use the gauges to verify the Delivery HGL of the pressure source.  If its different investigate why: Pressure Source operational settings, survey level inaccuracies etc.
    • If this is verified then go to sample sites the client has tested and measure static pressure.  If you get a different pressure, then there is a Client pressure instrumentation problem.  If you get a different HGL from the HGL you measured at the pressure source, then there is likely a network valving issue.
    • Ensure during this phase do it in times of day when background demand is not significantly drawing down the network HGL/Pressures.

    Having said that, I have never used models for the initial siting of ~80 PRV Sites for over ~50 pressure managed area designs.   These were selected primarily by network maps overlaid by topographical contours, streets, rail corridors and waterways.   I have used models to verify my design, but there were only a handful of cases that the model results indicated that I needed to move the PRV to somewhere else.

    Old-school but I would just get A1 or A0 plots with these layers, and take a highlighter pen and draw along the contour lines that below that height was above my maximum pressure target limit.   I would then take a different coloured highlighter and draw along the contour lines that above that height was below my minimum pressure target limit.  From this, with the highlighted contours and the constraints posed by waterways and street corridors, we could then easily start to draw where "natural" pressure zones started to form, where to site the PRVs at natural feed points near the maximum pressure contour, check that they weren't going to impact customers above the minimum pressure contour, and use waterways, railway lines and major roads as "natural" boundaries for these new zones.  We would then refine these by looking for ways to minimise the number of zone boundary valves to be shut or installed, and whether we could rationalise it more by installing some new bridging pipelines between the new zones to combine them into larger zones and reduce the number of PRVs required.  Optional but at the same time we would mark on areas or properties that we believed would be extremely sensitive to a pressure drop and try to on the maps look for ways to zone them outside the pressure managed areas.

    It was only after we finished maybe drawing for 4 hours on the A1s/A0s that we would then take the design and plug it into WaterCAD to check that it would work for Peak Hour minimum customer pressures, and fire-flow performance.  Once you do this a few times you get quite practiced at being able to mark up zone designs by eye.. Way faster method then trial and error in a model, at least for us! Wink

    Some warning though, hilly areas are really hard to do economical PRV zones in and are generally avoided in preference to better returns in flat, low-lying high pressure areas.  To height change in just one street means you may have some properties with minimum pressure at the top and high pressures at the bottom and one PRV may only be able to cover say 5-20 properties each, versus a preferable target of 1 PRV to 1,000-5,000 properties which is an expensive total investment once factored in the maintenance and operational costs that come with them.  A better strategy sometimes is to just use higher pressure class mains and PRVs on the plumbing connections!



  • M, what you really want to get from a hydrant flow test is the residual HGL ( pressure), in the distribution system. If you measure the HGL at the flowed hydrant, the value is  distorted by all the head loss in the hydrant lateral and barrel. Such residual pressures are not reflective the pressure in the main and are essentially useless for model calibration. The key results is the HGL in the system when the hydrant is flowing. You need two hydrants for that.

    If you are going to run a flow test and use it for calibration, do it correctly with a separate flowed and residual hydrant. See this blog I wrote about flow tests.

    https://blog.virtuosity.com/hydraulics-hydrology/what-good-is-a-single-hydrant-flow-test

    Ben's advice on setting up pressure zones is very good. What I like to do is establish "boundary contours". Anyone higher than that contour goes into the higher zone while anyone lower goes into the lower zone. Draw these contours on a map or GIS and include areas that are not already in your service area. This will come in handy later. The PRV should be located on or near these contour lines,.

  • Thank you both for  your very detailed answer. Just a couple of follow up questions. The client did not provide any pump status operating conditions when they ran their field tests. The system has two wells with pumps that pump directly to the distribution system and an existing storage tank that sits at a higher elevation but only has 10ft of depth.  When we are checking static pressures in the system (comparing field vs modelled results), what pump info do i need to input in the model. I assume if I run a hydraulics only steady state run, pump condtion is still  going to be very important? is that information that needs to be carefully documented in the field? 

  • Whether the pumps lifting from a Well to the distribution system's storage Tank are On or Off during field tests is often vital information.   Many zones this significantly alters the zone's static pressure distribution and their hydrant performance.

    Generally you will always need all boundary conditions to be simulated as accurately as possible, including background demands, control valve settings, and any sources of pressure supplying the distribution network.

    In order to design field tests, and often an engineer will need to provide a explicit test plan to a client or field team to ensure all important field variables are either controlled or at least measured,  I would normally look at a simulation of the predicted field test results with and without the lift pumps operating.  If there was a significant difference, then on the field test plan it would be specified what manual override settings needed to be used, either input via Remote Control or physically going to the Switchboard to pump the Pumps in Manual Off during the tests.   If this could not be controlled, then flow metering and pressure sensing on the pump station outlet was required to be measured simultaneously with field test measurements.   It is client dependent, but sometimes we needed to be a little blunt with the client in saying "We either test this properly using a test plan or we don't bother with testing at all" because calibration does need all significantly influencing variables to be measured reliably and accurately, whereas some clients will throw a bunch of data with you generated by ad-hoc testing with ad-hoc measuring equipment and it proves to be a case of Garbage in = Garbage Out.

    Similarly, even though tank is only 10 ft, should try to get this boundary condition measured as accurately as can if there is not already a tank level sensor + logger.  With today's technology, whilst in previous decades we might have had another person standing there to take readings who we talked to be phone or radio, now you can buy cheap loggers that can be installed at the boundary head points temporarily and will send the remote pressure data to the Cloud  (Eg. We use i2OWater pressure sensors now for this plugged into the system with hydrant caps + pressure port spigot, reducing the amount of people needed for testing)



  • To supplement Tom and Ben's answers regarding hydrant flow test calibration, please see the following article and its imbedded video: Calibrating a model based on hydrant flow tests


    Regards,

    Jesse Dringoli
    Technical Support Manager, OpenFlows
    Bentley Communities Site Administrator
    Bentley Systems, Inc.