How do I measure tank gauging accuracy?

Dec. 21, 2005

ControlGlobal.com presents "Ask the Experts," a column moderated by noted process control authority and columnist Bela Liptak. Save yourself the hefty consulting fees by getting the answer to your question from Liptak's cadre of professional automation experts.

QUESTION:

I UNDERSTAND that the core of hydrostatic tank gauging (HTG) mass measurement is simply a d/p gauge and a characterizer. My question is this: If I profile a tank with water to develop the characterizer profile, will the mass readings be accurate for varying specific gravities? Are there inherent limitations either in theory or practice using this method? If I am using a high-quality d/p gauge with an accuracy specification of 0.1%, what could I expect the overall system accuracy to be, and where can I most likely expect to see measurement errors? To better describe my situation, our typical tank is 9,000 gallons with a 190-in. straight side. Our storage tanks see product specific gravity variations from 1.05 to 1.15 sgu. Can I expect valid readings with this kind of density change? Thanks for your help. I really enjoy reading you column, and have learned much from it.

Jim Laning, Automation Engineer, Genencor International

ANSWER:

FIGURE 1: PRESSURIZED, SPHERICAL TANK

Pressure is detected at three elevations in a pressurized, spherical tank.

Detect Pressure at Three Elevations
Very good question. In a typical 345-ft diameter, 750,000-barrel, American Petroleum Institute (API) storage tank, it takes 8,000 gallons to raise the tanks level by 1 in. Therefore, if level measurement is used as the basis for custody transfer, an error of 1 in. produces an error of 8,000 gallons. A 0.1% d/p gauge will have that error, if its span is 1,000 in. of H2O, which is also known as 83 ft of head. On tall tanks, to reduce the spans, we use multiple d/p gauges. Level detection based on d/p can be corrected for density variation, if the density in the tank is uniform.

The installation is important. As illustrated in Figure 1 for a pressurized spherical tank, we detect pressure at three elevations: P1 at the bottom, P2 near the center, but always below the P3 level at the top. In this way, P1 - P2 divided by the distance between them gives density, and P1 - P3 divided by density gives level. This is true only if: 1) the specific gravity is uniform throughout the tank (lighter fluids tend to rise); and 2) you have temperature compensation.

This level must be entered into your "strapping" table for the particular tank to obtain the actual level. This table must be updated yearly because such large tanks do change shape with time. The water level below the organic also must be entered into the strapping table because the accumulation of water at the tank bottom does change.

With a 0-200 in. span, the error of 0.1% corresponds to 0.2 in. The error in the density correction should not increase that error to more than 0.4 in.

The usual problem is locating the pressure tap at P2, so that it will always be below the level in the tank, but still sufficiently higher than P1 to give you a good span for density detection (>20 in). The other usual problems are that temperature and/or density varies with elevation in the tank.

Radar is an alternate to d/p detection, but thermal tank expansion and roof deflection must be corrected for. As I'm sure you know, dip tapes are also used with their corresponding potentials for human errors.

Béla G. Lipták, PE , CONTROL columnist