The valves of the prover should be kept free from dust and well greased at all times. Any grit in the valves will soon scratch the bearing surfaces and cause leaks. When not in use, the circular slidevalve should be covered to keep out dust.

3.2 Prover Calibration

There are two basic methods of calibrating a bell prover, "bottling" or "strapping" [3, 4]. "Bottling" a prover involves transferring a cubic foot of air between a cubic foot bottle or portable standard and the prover.

"Strapping" a prover is a method of calibration in which the physical dimensions of the bell, the tank, and the levels of the sealing liquid are measured and the capacity is then computed.

Dents in the portion of the bell that moves into and out of the sealing oil should not exceed 1 cu in per cubic foot of rated prover capacity. The maximum diameter of a given cross section of the bell should not be more than 1 percent greater than the average diameter of the cross-section [5].

The maximum temperature difference of all equipment including prover, cubic foot bottle, standard sealing fluid, and ambient air should not exceed 0.05 °F and should be held constant throughout the calibration period [5].

a. Cubic Foot Bottle

A cubic foot bottle such as would be used for bottling a prover is illustrated in figure 4. The nickel or chromium-plated copper bottle is open at the end of the lower neck and may be lowered into a tank of sealing liquid, similar to prover oil. As the liquid enters at the bottom and fills the interior up to a graduation mark on the gage glass in the upper neck. 1 cu ft of air is displaced and flows through a connecting tube to the prover bell, which rises.

The transfer of air can be from the descending bottle to the rising bell of the prover, or from the descending bell to the rising bottle. In either case the results should be approximately the same, except for slight differences in the behavior of the sealing liquid. A residual amount will remain on the walls of the instrument that has been standing in a raised position before it descends, while the rising surfaces of the other will carry adhering liquid upward. Because the total involved surface area of the bell will not be the same as that of the bottle, a perfect duplication of performance between the two cases cannot be expected.

Transfers can be made with the system under normal prover pressure, or with the counterweights adjusted to balance the bell at atmospheric pressure. The first method produces the slight difficulty of setting the liquid level in the gage glass of the bottle while it is slightly below the level of the surface in the tank; also it involves the necessity of raising the bottle at the end of the transfer until one or two small bubbles of air are seen to escape.

FIGURE 4. Exterior view of a cubic foot bottle.

For each cubic foot of air to be passed into or out of a prover, the bell is first placed in a position, by manipulating the rotary slide valve, so that the trial begins with a cubic-foot mark on the scale opposite the pointer. If the scale, the standard, and the procedure were all perfect, the next cubic-foot mark would be exactly opposite the pointer at the end of the test. The difference between 1 cu ft and the volume indicated by the scale is the erro determined by the calibration. This procedure is repeated to measure successive cubic-foot increments until the entire range of the prover has been tested.

Frequently another device is used in place of the bottle called a "Stillman cubic-foot standard" (fig. 5) after its inventor, the late Mr. M. H. Stillman of the National Bureau of Standards [6]. A movable bell rises and descends in an annular tank very similar to that of a prover. The bell is guided by a central vertical column sliding in a close-fitting cylinder, and upward travel is limited by an adjustable stop. This standard is about as accurate as a cubic-foot bottle, is used in a similar procedure, and

FIGURE 5. Exterior view of a portable cubic foot standard.

offers the additional advantage that it is easily portable. By draining the sealing oil from it, and placing it in its carrying case, a task which can be accomplished quite readily, the user can easily transport it from place to place, say by automobile. It is as easily set up and placed in service, except that a period of several hours is sometimes required for the temperature of the standard to stabilize and become equal to that of the prover to be tested. Because of its portability and its relatively small size compared to a bottle in an immersion tank, it has found wide-spread acceptance by State public utility commissions and weights and measures offices. Like the bottle, it can be sent to NBS for calibration.

Temperature and barometric pressure are very critical in "bottling" procedures. A high degree of stability of these two factors is necessary.

This brief description of calibration by "bottling" is considered sufficient for the purposes of this handbook. Complete detailed instructions on this tradi tional and widely used procedure are given in publications [3] and [7] in this list of references.

b. Strapping

The procedure of "strapping" is illustrated by figures 6a and b, which show diagrammatically the cross section of a typical bell-type prover. The capacity of the bell, i.e., its internal volume, plus the volume of the metal in the bell, is designated as the "outside volume." Each term applies only to the volume above the liquid surface, or "seal."

The validity of the method is based on this relationship: When a prover bell is lowered from any position to any other position, the volume of gas discharged will be equal to the outside volume above the seal at the first position, minus the outside volume above the seal at the second position, plus the volume of the metal in the scale that becomes immersed, and minus the volume of the fluid that rises between the outside of the bell and the main tank. The truth of this statement is not immediately obvious, but will be made clear by a simple analysis. First, the following must be defined:

Q=Volume of gas discharged by prover bell. V=Change in exterior volume of the bell above the liquid surface.

B = Change in interior volume of the bell above

the liquid surface.

W = Volume of liquid rise between inner tank and interior surface of the bell caused by liquid displaced by metal in bell and scale. M = Volume of metal that becomes immersed in liquid.

S= Volume of metal scale buttons, guide rods, and thermometers that becomes immersed in liquid.

T= Volume of liquid rise between main tank and the outside of the bell caused by metal in bell and scale.

Figure 6a illustrates the bell position at 0 on the scale, with an internal pressure represented by the difference in levels of the sealing liquid inside and outside of the bell, designated by H. As the bell descends to position 100, as shown in figure 6b, if there were no change in the level of the sealing liquid, the volume of gas, Q, discharged would be equal just to the change of interior volume of the bell, B, above the liquid surface. However, the level of the surface in the annular space between the main tank and the outside of the bell, and the level between the inside of the bell and the center tank, will rise because some liquid is displaced by the metal in the bell and the scale. The volume of liquid which rises between the inner tank and the interior surface of the bell, W, displaces gas and therefore adds to the volume of gas, Q, which is discharged. Thus,

Equation (6) is the principle upon which the method is based.

Adjust the prover oil level so that the surface of the oil remains within a constant-diameter section of the outer tank throughout the travel of the bell. Measure the outside diameter of the bell. Mark the bell at the oil level with the bell positioned so that the pointer is opposite the scale markings corresponding to 10, 30, 50, 70, and 90 percent of the full scale reading. Measure around the circumference of the bell with a steel tape, preferably calibrated for direct reading of outside diameter, at the plane of each of the marks on the bell. Use the average of these five readings as the outside diameter. (The diameter may also be determined by measuring the circumference with a steel tape capable of being read to the nearest 0.01 in. Calculate the diameter by dividing the average circumference by 3.1416 then subtracting the thickness of the tape.) The surface of the bell should be wiped off before any measurements are made.

Manufacturers state that prover bells up to 10 cu ft are usually quite cylindrical, with circumferences which do not deviate more than about

plus or minus 0.03 in from the average circumference. If the deviation from the average circumference does not exceed ± 0.0625 in, the calibration may be based on the average girth, and the scale would have equal division throughout its length. However, if the bell deviates more than ±0.0625 in from the average circumference, the strapping should be subdivided to volume increments of essentially constant girth. In laying out a plan for a calibration, the scale point at the middle of the increments being tested should be selected, and the corresponding plane of intersection of the liquid with the bell should be observed. The circumference of the bell should be measured in this plane.

The volume of liquid which rises in the tank is computed from the change in surface level and the bell and tank dimensions. Liquid-level reading should be carefully made by means of micrometer depth gage held firmly on a flat upper surface of the tank.

For the relationship Q=V+S-T to be true, two requirements must be satisfied. First, the prover and counterweights must be adjusted to insure that the pressure within the bell is the same at any position as well as those seen in figures 6a and b. This

may be checked by connecting a good draft gage or inclined manometer to indicate the internal pressure. Another procedure is to open the interior to atmosphere by means of the prover rotary valve, adjust the counterweights until the bell remains stationary at any position, and perform the calibration under these conditions. Second, the effect of oil drainage down the walls of the bell on the surface level of the liquid should be negligible. This generally does not affect the surface level of the liquid after the bell has been draining for several minutes. A typical set of measured values and the calculations required are shown below [3]. The steel scales, verniers, and micrometers used in strapping should be of high quality; the bias or error inherent in the instruments should be relatively insignificant compared to the uncertainties associated with making the measurements. If reasonable care is used in calibrating, say, a 5 ft3 prover, the maximum variations between measurements by different observers and at different times should be not more than: 0.01 in for the length of the scale; 0.001 in for the average thickness and width of the scale; 0.001 in for the change in fluid surface level; and 0.003 in for the bell diameter and the distance from bell to tank [4].

4. Inspection of Commercial Devices

For a discussion of the purpose and scope of "inspection" as distinguished from "testing," see Section 4, National Bureau of Standards Handbook 44, Specifications, Tolerances, and Other Technical Requirements for Commercial Weighing and Measuring Devices.

Particular attention should be paid to code requirements pertaining to units of primary indicating elements and corrections for altitude. An altitude correction table for ranges of elevation (customary units) where the devices may be installed is provided in table 1. It would also be advisable to verify that, if the altitude correction factor is other than 1.00, it is used in computing the billing for commercial transactions made with the device. In addition, an altitude correction table (table 2) is provided for ranges of elevation in metric units.

TABLE 1. Altitude corrections factors (customary) [8, 9]

5.1. Prover Room [3]

The selection of a proper location for a prover room requires careful study and attention to surrounding conditions. An interior well-lighted room (one without outside walls or windows) is preferred. Direct sunlight should never be allowed to fall on the prover or meters under test. Care should be exercised in heating or cooling the room. Temperature variations of 5 °F may be attained in inferior installations. In no case should the prover be located near radiators, steam pipes, or hot and cold air ducts. In addition, temperature changes greater than 5 °F should not occur during a period of 24 hours, since it is essential that the prover and the meters to be tested be as near the same temperature as possible. A difference of 5 °F between the temperature of the prover and the temperature of the meters under test may produce an error of 1 percent.

Air conditioning for a proving room should not be confused with ordinary space cooling as generally employed to achieve comfort for the room occupants. Ordinary "commercial" air conditioning can be worse than no cooling at all, since it can result in a fairly rapid cycling of temperatures and variations in temperature throughout the room. This could cause erratic results of greater magnitude than would be encountered if the temperature were allowed to drift throughout the daybut with lesser momentary temperature discrepancies at vital points.