Problem with guided wave radar level measurement in euro-diesel storage tanks - ROSEMOUNT 5300

[vjaksic81]

Dear colleagues,

I'm having serious troubles for more then 6 months trying to measure level of euro-diesel fuel in cylindrical, with fixed conical roof and thermal isolated mantle storage tank. Roof is not isolated. Tank dimensions are: approximately height 15000 mm, diameter 21700 mm, usable storage volume 5000 m3. Tank is not pressurized.

I'm trying to set up GWR Rosemount 5301 level measurement system for our client inventory system and to harmonize it (as much as I can) with custody level measurement.

Custody level measurements are performed by accredited measuring company with calibrated and certified steel measuring tape with weight (accuracy is +/-(0.3+0.2*L) mm where L is in meters). Custody level measurements are performed by veering down measuring tape trough measuring opening on roof of the tank, and submerging it in fluid until weight hit lowest measuring point (steel plate at 160 mm from bottom of the tank). After that, tape is pulled up to read level of fluid witch leaves distinctive mark on tape.

GWR is set up at revision opening (24" wide) on the other side of the roof from custody level measurement place. Radar works on TDR method, and it is version with 4 mm SS single cable with SS weight on the end with accuracy of +/- 3 mm. Installation of device were performed according to all manufactures mounting considerations given in reference manual (distance from wall of the tank, distance from obstacles in the tank, free space around probe...) Device is not installed in stilling pipe, it freely hangs from revision opening downward.

After setting up device trough Rosemount radar HART configuration software (probe length, tank height, lower measuring range, upper measuring range, damp time, measuring place height offset from custody measurement place etc...) Radar level measurement wasn't in accordance with measured custody level (all tough it should have been, because we took height offset between custody and radar measurement place quite accurately). So I did tank height correction in radar to get it in accordance with measuring tape, and then started problems...

When ever level in tank is near level where radar where harmonized with tape (L +/- 0,5 m), radar measures correctly and is in +/- 3 mm range from level measured with measuring tape (witch is in accordance for that instrument and its calibration list provided by manufacturer), but when level of fluid rises radar starts to measure lower level then tape, and the absolute difference between two measurement is continuously growing as level of fluid rises more and more from level where 2 measurements were harmonized (L + 2340 mm diff is -10 mm, L+3595 mm diff is -15 mm etc...) Now, when level in tank starts to decrease and come close to harmonized level L, difference between two measurement comes in reasonable range. When level decrease from harmonized L, difference starts to occurs again, but this time in different direction, now radar measures more than tape and difference continuously rises as level of fluid decreases (Example L-1 m diff +9 mm, L-2.3 m diff is +13 mm...) When level in tank goes up and once again come near to level where 2 measurements were harmonized difference between measurements comes in expected range).

This happens all the time on 2 equivalent tanks with same equipment and same measuring methods and places. Before Rosemount 5300 series radar, I've tried KROHNE OPTIFLEX 2200C which were performing same as described before... We also tried to harmonize 2 measurements on different levels L, but behavior was same... near L everything is fine, as soon as level rises or fall for more then 1 m from L, absolute difference between 2 measurement continuously grows with distance from L.

If any one have any idea what could be going on I would be forever grateful...

Best regards,

vjaksic81

 

[danw2]

Is the error you 'see' an indicated level value in the device's display, or are you reading an interpreted value from the 4-20mA output in a DCS/PLC/indicator/whatever?

Which parameter are you changing to 'harmonize' the TDR with your reference?

 

[xnuke]

There's definitely a span error somewhere, either in the calibration of the unit or in the scaling of your display, which is why danw2 asked if your reading is from a display on the device (indicating a span error in the unit's calibration) or in a DCS/PLC/indicator (indicating a scaling error in the readout device).

Could you please describe how you calibrated the unit once it was installed?

xnuke
"Live and act within the limit of your knowledge and keep expanding it to the limit of your life." Ayn Rand, Atlas Shrugged.
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[vjaksic81]

@danw2 error I see is indicated level value read from the device when I connect to it with Emerson 375 Hart communicator or via Rosemount configuration software and Viator Hart modem on my field notebook. When I tried KROHNE OPTIFLEX 2200C that error was also visible on device display. Emerson supplier, supplied me with me Rosemount 5301 without local display.

I've trimed "Tank Height" parameter to harmonize radar with measurement tape, because radar measures distance from his starting point to surface of fluid, and then subtract that distance from "Tank Height" to gain level. I have also tried to do distance calibration (in example for given Tank Height i trim distance to achive tape measured level)

Readings on SCADA are readings from radar +/- 1-2 mm (given to A/D conversion of PLC AI, and stability of 4-20 mA signal from device) I've calibrated analog input of PLC using device posibility to set desired mA at its output. For 4 mA system displayed 1mm, for 12 mA system displayed half of span +/-1 mm, for 20 mA system displayed full span +/-1 mm, so I have no doubts that PLC and SCADA are working good enough.

@xnuke once the device was installed, I've measured and entered "length of probe", then measured tank height from the begining of probe to bottom of the tank (using sam tape used for custody measurement), and entered that value as "Tank Height" parameter, after that I've set up level value for 4 and 20 mA signal (according to set up in DCS), and put device in operation. Than investor did custody level measurement and we compared those 2 meaasurements and there were some difference so I've cahnged "Tank Height" to compensate this difference. Behaviour of devices after that I've described in my first post.

 

[danw2]

I'm mystified. Changing the tank height amounts to an offset adjustment and should not affect the slope of the calibration line, because the level is calculated as tank height minus distance to surface so a change of tank height is constant for all measurements over the span.

If there's a slope/span adjustment, like Magnetrol's factory parameter called "CONVERSION_FACTOR: the slope of the factory-set calibration line", or Siemens "Conversion Factor (Conv Fct) (factory authorized personnel only) Default: 985, Calibration parameter. This value equals (mm/tick)*2048. The nominal value is 963. This is the slope of the straight line measurement equation; factory password required", I can't find it.

However, you haven't mentioned doing the first part of a distance calibration as described in the manual.

Have you done that? I'm not sure the procedure will skew the slope, but one works with what one has available.

Did the Krohne produce erroneous readings in the same fashion?

 

[vjaksic81]

@danw2 Couple days ago I've done distance calibration as you have described in yours reply (folowing instructions from page 7-17 Rosemount reference manual 00809-0100-4560 rev BA july 2009).

Before that I was doing level calibration as described in same chapter. We've measured level in tank with custody tape and adjust tank height parameter to gain same level from radar. That adjustment we've done at level of 7354 mm.

After that level of fluid have been decreasing for few days and radar produced errors as described in previous posts (for level 6456 radar measured 6476 (+11), for 6364 radar measured 6372 (+8), for 5419 radar measured 5432 (+13).

When we came to the tanks to see wtf is going on, custody measurement was 7861 mm, and radar measured 7859 mm (as I've described: when fluid level reaches zone near adjustment level (7354 in this case) radar level measurements are in order. So every comparison of measurements was uselles...

Radar on second tank was adjusted (with Tank height parameter) at 4555 mm, after that level raised custody measured level was 8150 mm and radar measured 8135 (-15 mm), then level started to fall... (for 7807 - radar measured 7795 (-12 mm), for 7044 - radar measured 7034 (-10 mm),for 6493 - radar measured 6483 (-10 mm), for 5400 - radar measured 5396 (-4 mm). As you can see, as the level approaches adjusted level of 4555 mm error decreases... then level raised to 8652 mm and radar (with same parameters) measured 8633 (-19 mm), after that we've measured 8080 and radar measured 8063 (-17 mm), than 7252, and radar measured 7242 (-10 mm) etc... error behavior is repeating...

Krohne produced erroneous readings in same fashion... And we always had product surface as calm it can be in tank when doing radar set up...

I'm mystified for 6 months

 

[ccfowler]

Are there any significant temperature variations?

Valuable advice from a professor many years ago: First, design for graceful failure. Everything we build will eventually fail, so we must strive to avoid injuries or secondary damage when that failure occurs. Only then can practicality and economics be properly considered.

 

[vjaksic81]

Tanks are in open field, and under various weather conditions. Tanks sides are insulated so temperature of fluid inside tanks changes slightly during day. Amount of fluid in tanks is such that its immpossible for significant temperature changes to occur. Behaviour of measurements was same from February when we install first Krohne Optiflex 2200C radars (when air temperatures were around 5°C) untill now when we have air temperatures of 30+ °C... How ever roof of the tanks are not insulated, so I've assumed (at one momment) that due to the heating of the tank roof, roof may elevate, and slightly move radar reference point up, which should increase measured distance from radar reference point to the fluid surface, and, due to the constant TH parameter, lower level then acctual... But radar measurement behaviour is as described in previous posts under all weather circumstances, and independent of time of the day...

 

[ccfowler]

Apparently, temperature is not a likely problem.

Your sites are obviously scattered over a significant area, so just how well regulated and consistent is the quality of the electric power supply to the various instruments? This may seem to be a silly question, but a long time ago, I had an experimental instrument system in a relatively remote location that was being severely influenced by surprisingly wide variations in the AC power supply. After seemingly crazy instrument indications, I installed a recorder to monitor the voltage of the AC supply to the instrument package, and the duration and severity of variations (mostly extremely low) fully explained the problem. Once that was cleaned up, all was well. This may be worth a careful look for your situation.

For what it is worth, I had been repeatedly assured that the voltage fluctuations that I recorded could never happen. The printed record obviously proved that the impossible was not impossible after all.

Valuable advice from a professor many years ago: First, design for graceful failure. Everything we build will eventually fail, so we must strive to avoid injuries or secondary damage when that failure occurs. Only then can practicality and economics be properly considered.

 

[hacksaw]

the negative off-set in radar level compared to the physical inventory, appears to be due to a change in the dielectric constant in the vapor space as the tank is filled. This is more noticeable when you are filling the tank in the upper third of the tank.

The amount of change in the vapor space dielectric constant is really small on the order of 0.1% relative to that of dry gas, so you might look at the blanketing gas flows

 

[hacksaw]

you need to enable the DVC option available with the 5300, it is normally reserved for dense vapors but the need is driven in part by your measurement range and the need for high resolution measurements.

 

[vjaksic81]

@hacksaw interesting... I'll definitely try that, hope that it will do the job..

 

[vjaksic81]

DVC option comes only with 4U or 4V type of probes (rigid single lead 8 or 13 mm for vapors), mine are 5A flexible single lead with weight...

 

[hacksaw]

you are up against the limitations of their measurement, the DVC correction can only reduce the vapor space error not eliminate it, and you are already pushing the limits of the TDR measurement in a dry gas.

welcome to the world of precision measurements

 

[danw2]

Sounds like hacksaw has identified the cause.

Is the error skew repeatable enough to build a look-up table in the DCS to correct for the deiviation?

 

[hacksaw]

eye-opening issue stumbled upon by the o/p