9th ISFFM Arlington, Virginia, April 14 to 17, 2015

NIST’s Fully Dynamic Gravimetric Liquid Flowmeter Standard

Jodie G. Pope1i, Aaron N. Johnson1, James B. Filla2, Joey T. Boyd1, Christopher J. Crowley1,
and Vern E. Bean1
1
Fluid Metrology Group, Sensor Science Division, National Institute of Standards and
Technology, Gaithersburg, MD
2
Office of Information Systems Management, Innovations and Solutions Division, National
Institute of Standards and Technology, Boulder, CO

Abstract
We describe a new dynamic, gravimetric, liquid flow standard (LFS) that determines flow by
measuring the rate of change of the liquid mass accumulating in a collection tank. The LFS is a
fully- automated,15 kg/s system that uses a proportional–integral–derivative (PID) control loop
to achieve liquid flows with a stability that is the smaller of 0.001 % or 0.1 kg/s. The expanded
uncertainty (corresponding to 95 % confidence level) is 0.021 % for the flow range of the
standard; 15 kg/s to 0.22 kg/s. We verify the uncertainty budget by comparing the LFS results
with two well-established NIST primary flow standards.

1. Introduction
Dynamic weighing methods are well documented and used by national metrology institutes
to calibrate liquid flow meters [1,2]. These methods can achieve low uncertainties while
avoiding the expense and complexity of either a flow diverter (generally required by static
weighing methods) or the mechanical complexity of provers. Here, we describe the National
Institute of Standards and Technology’s (NIST’s) new 15 kg/s liquid flow standard (LFS). This
primary standard uses dynamic weighing to measure flows ranging from 15 kg/s down to
0.22 kg/s with an expanded uncertainty (corresponding to 95 % confidence level) of 0.021 %.
The 15 kg/s LFS is a fully-automated, closed-loop system comprised of three major
components: 1) a flow generation and control system, 2) a dynamic weighing system that
makes SI (International System of Units) traceable flow measurements, and 3) a test section
where customer meters-under-test (MUT) are installed and calibrated. The dynamic weighing
system consists of a weigh scale and a collection tank. The scale readings are recorded at
approximately 0.2 s intervals during flow accumulation into the collection tank. As shown by
Shinder and Moldover the slope of the time-stamped, buoyancy-corrected weight
measurements equals the mass flow [3].
The uncertainty of the 15 kg/s LFS is determined using the Monte Carlo method [4]. (See
Section 6.) We verify the calculated uncertainty by comparing the 15 kg/s LFS results with two
well-characterized NIST flow facilities, the 65 kg/s primary LFS [5] and 2.5 L/s primary piston
prover [6]. Figure 1 shows comparison results, which are in good agreement, being less than
0.031 %; Section 5 explains this comparison in detail.
This manuscript describes the 15 kg/s LFS, documents its capabilities as determined by
comparisons with well-established flow standards, explains the principle of the dynamic
weighing method, and shows the results of the Monte Carlo simulations.

                                                            
i
Corresponding Author: [email protected]

1 
 
9th ISFFM Arlington, Virginia, April 14 to 17, 2015

Fig 1. Comparison of new 15 kg/s with NIST’s 65 kg/s and 2.5 L/s liquid flow standards using
5 cm transfer standard (solid symbols) and 2.54 cm transfer standard (open symbols). The pink
and the blue base lines are the mean reading of the 2.54 cm and the 5 cm transfer standard on
the 15 kg/s LFS, respectively. Error bars show the expanded uncertainty in the measurement.

2. Description of the Liquid Flow Standard

The 15 kg/s LFS is a closed-loop liquid flow calibration facility that is automated using the
National Instruments LabViewii environment. Custom software was developed to operate
mechanical components of the standard and to acquire and store calibration data. Calibrations
are done at ambient temperatures with the fluid pressurized to no less than 140 kPa to avoid
cavitation. The schematic in Fig. 2 shows the major components of the LFS including: 1) the
flow generation and control system consisting of a variable flow pump, reservoir tank, check
standard (or reference) flowmeter, butterfly valve, and data acquisition system with digital
proportional-integral-derivative (PID) controller; 2) a test section that accommodates a MUT with
pipe diameter ranging from 1.25 cm to 5 cm; and 3) the dynamic weighing system comprised of
a collection tank and weigh scale. A picture of the 15 kg/s LFS is shown in Fig. 3. Table 1
gives the nominal characteristics of the LFS.

2.1 Flow generation, control, and stabilization

The 15 kg/s LFS generates and maintains steady flow at a desired set point using a pump, a
butterfly valve, and reference flow meter. The pump pressurizes and circulates the reservoir
tank water throughout the system. The flow and line pressure are stabilized at a desired set
point using a PID controller to continuously adjust the pump speed and butterfly valve. When
stable, the set point specified in the LabView program will equal the flow indicated by the
reference meter. The flow is considered stable if for a time period of 60 seconds, the maximum
of the first and second derivative of the flow from the reference meter is less than 0.0001 kg/s2
and 0.01 kg/s3 respectively and one of the following conditions are met: 1) the maximum percent
error in the flow from the set flow is less than 0.001 % or 2) the maximum flow error from the set
flow is less than 0.1 kg/s.
                                                            
ii
Certain commercial equipment, instruments, or materials are identified in this paper to foster understanding. Such identification
does not imply recommendation or endorsement by the National Institute of Standards and Technology, nor does it imply that the
materials or equipment identified are necessarily the best available for the purpose.

2 
 
9th ISFFM Arlington, Virginia, April 14 to 17, 2015

Fig. 2. Schematic of the 15 kg/s LFS. During a calibration, the flow is stabilized after closing
the bypass valve and diverting the flow through the collection tank into the reservoir tank.

Fig. 3. The 15 kg/s LFS.

3 
 
9th ISFFM Arlington, Virginia, April 14 to 17, 2015

Table 1. Nominal Characteristics of the 15 kg/s LFS

water(currently), 5 % propylene
Fluid(s)
glycol in water (future)
Flow range [kg/s] 0.22 to 15
iii ◦
Temperature Range [ C] 20 to 25
Pressure Range [kPa] 140 to 550
Volume of collection tank [m3] 0.90
Volume of reservoir tank [m3] 1.1
Diameter of piping [m] 0.051
Length of test section [m] 3.3
Connecting volume between MUT and collection
0.01
tank [m3]
Collection times 15 s to 3000 s

During the flow stabilization period the system operates in an open loop mode, where the
bypass valve is closed and both the tank valve and the dump valve in the collection tank remain
open. Flow leaving the reservoir tank is directed through the reference meter, the MUT, and
finally through the collection tank before it is returns to the reservoir tank. The dump valve
closes to start a flow measurement. Flow accumulates in the collection tank for 15 s to 200 s
depending on flow. At the lowest flow of 0.22 kg/s the tank can accumulate liquid for up to
3000 s, however, this amount of mass is not necessary to acquire a calibration point and the
collection is limited to 200 s. Following the flow measurement interval, the dump valve opens to
drain the collection tank in preparation for the next flow measurement. Therefore, the collection
tank is constantly either filling or draining during a calibration.
An alternative mode of operation of the 15 kg/s LFS is to use the reference meter as the
working standard to calibrate customer meters. This mode also enables research on the effects
of unstable flows on meter performance. In this mode, the LFS collection tank is bypassed.
The system tank valve is closed and the bypass valve opened to achieve a flow loop.

2.2 Test section

The test section is a straight length of 5 cm diameter stainless steel piping. Flow calibrations
are performed with the MUT installed after a minimum of 30 diameters of straight pipe length.
We measure temperature and pressure upstream and downstream of the MUT. The fluid
temperature is not controlled and increases by as much as 0.5 K due to flow work from the
pump during a flow collection. These temperature changes cause the mass of liquid stored in
the connecting volume (i.e., volume between the MUT and pipe exit leading to collection tank in
Fig. 2) to change between the beginning and ending of a collection. These mass storage
effects are accounted for in the mass flow determination and uncertainty analysis of the LFS.
The pressure in the test section is controlled and remains essentially constant during a
collection, but can vary with flow and the size of the MUT. In general, the pressure increases
for larger flow and smaller flow meters. During a calibration we make sure the pressure is
sufficiently large to avoid cavitation. During the validation of the 15 kg/s LFS, the pressure in
the test section was highest when calibrating a 2.5 cm coriolis meter; it ranged from 140 kPa at
a flow of 0.22 kg/s to 445 kPa at a flow of 5 kg/s. When calibrating a 5 cm coriolis meter, the
test section pressure was nominally 245 kPa at a flow of 15 kg/s. The maximum pressure is
limited to the pump output; 552 kPa.

                                                            
iii
Calibrations are performed with the fluid at the ambient temperature.

4 
 
9th ISFFM Arlington, Virginia, April 14 to 17, 2015

2.3 Dynamic weighing system

During a primary flow calibration the bypass remains closed and the flow is continuously
either filling or being drained from the collection tank. The tank outlet is a 10 cm hole that is
opened and closed by the dump valve (Figs. 2 and 3). The tank drains directly into the reservoir
tank, which is suspended above the weigh scale, but not in physical contact with it. The
collection tank is covered to prevent water from splashing out of the tank. The inlet plumbing
into the collection tank terminates with a cone that lies beneath the tank cover. The cone and
cover are shown in Fig. 4. The collection tank including the cone and cover are the only parts of
the standard in contact with the scale. The top and bottom portions of the cone form an annulus
that direct flow into the collection tank. The cone diffuses any horizontal component of
momentum with radial symmetry, thereby, minimizing flow-dependent horizontal forces on the
weigh scale.
The weigh scale has a 1.2 m x 1.5 m weigh platform with a 1500 kg capacity and 20 g
resolution. The data acquisition system continuously reads the scale output that is updated
every 0.209718 s ± 0.000001 s, based on the scale internal 24 MHz clock. The scale timing
accuracy was verified by multiple long term timing tests performed on the scale. Measurements
from the meters and line temperature and pressure sensors are taken every 0.2 s, and from the
environmental monitor every 0.46 s. The software uses the scale time as the reference for all
other measurements.

Fig. 4. The cover and cone removed from the collection tank. The cover rests on the
collection tank without touching the cone that is used to direct the liquid into the tank. The
incoming liquid flows between the two plates that make up the cone.

5 
 
9th ISFFM Arlington, Virginia, April 14 to 17, 2015

3. Flow Measurement Principle of Dynamic Weighing

Flow determinations are based on the rate of change of the buoyancy corrected mass in the
collection tank. The weigh scale reads the sum of three terms: 1) the weight of the fluid, 2) the
tare, and 3) the momentum exchange of the fluid at the point of impact in the collection tank:

tare, (1)

where W is the calibrated weight reported by the scale, is the acceleration due to gravity, and
1 ⁄ is the apparent mass. The apparent mass equals the true mass
multiplied by a buoyancy correction factor where and are densities of air and water
respectively. The vertical momentum transfer from the falling liquid to the scale equals the
mass flow multiplied by the vertical velocity, . Here is the mass flow we are trying to
measure and is the velocity at time at the interface where the jet impinges with the water in
the collection tank.
Figure 5 shows a control volume inside the collection tank as it is being filled. The tank on
the left shows the state of the system at time t1, which precedes the tank on the right that shows
the state of the system at time t2. Based on Eqn. 1 the difference in the scale weights between
and is

1 ⁄ . (2)

A mass balance around the control volume in Fig. 5 results in the following expression

, (3)

where the first term on the right hand side is the mass accumulated into the control volume
during the collection period, and is the mass of the annular jet falling into the collection
tank.

Fig. 5: The red dashed line and the tank boundary makes up the control volume. The tank
on the left is the state of the control volume at t1 and the tank on the right is the state of the
control volume at t2. Drawing is for illustration purposes only.

6 
 
9th ISFFM Arlington, Virginia, April 14 to 17, 2015

Under steady flow conditions the mass of jet is

jet , (4)

where is the time it takes the jet to fall from height to height . Neglecting air
resistance the fall time is determined by the following kinematic equation

⁄ 1 ⁄ (5)

where and are the respective vertical fluid velocity at and . By combining Eqns. 2
through 5 the change in the scale weight between and is

1 ⁄ . (6)

Solving Eqn. 6 for mass flow gives

, (7)
⁄

where / 1 air ⁄ liquid is the buoyancy corrected, calibrated scale mass readings.
For infinitesimally small times between scales readings, Eqn. 7 can be expressed in differential
form by

LFS , (8)

where the subscript “LFS” has been added to the mass flow to denote that it is determined by
the LFS dynamic weigh standard. The simple expression in Eqn. 8 relates the mass flow into
the collection tank to the rate of change of the buoyancy corrected, calibrated scale readings.
This formulation is valid for steady flows with no mass storage in the collection tank. As
discussed by Shinder and Moldover, the model must be extended to apply it to unsteady
flows [3]. However, the flow control loop used in the LFS primary standard is able to provide
suitably stable flows so that Eqn. 8 is used as the basis for measuring mass flow.
Because the scale response is linear over its operating range as determined by repeated
calibrations and the flow is stable, the scale mass versus time is a straight line with the slope of
the line equal to the mass flow. Herein, we use linear regression of the time stamped scale
readings to determine the slope, and hence the mass flow. The linear regression model for the
slope is [4]

∑ ∑ ∑
LFS , (9)
∑ ∑

where and are the respective time and mass reported by the scale. We also determine the
intercept:

7 
 
9th ISFFM Arlington, Virginia, April 14 to 17, 2015

∑ ∑ ∑ ∑
c , (10)
∑ ∑

which is used for the uncertainty analysis. The masses ( ) and times ( ) in Eqns. 9 and 10
are boxcar filtered to attenuate noise in the data [7]. The boxcar averaged times are

1
̃ (11)

and the boxcar averaged masses are

1
, (12)

where ̃ and are the respective unfiltered times and masses, and is the size of the boxcar
filter. The length of the mass (or time) arrays of the boxcar averaged data used to calculate
slope is 1 where is the length of the unfiltered data.

3.1 Flow at the MUT

The mass storage in the connecting volume between the MUT and the collection tank must be
accounted for to compare the mass flow at the MUT to the standard. The fluid temperature and
pressure is measured in the connecting volume at 0.2 s intervals. The measured temperatures
and pressures are used to calculate the fluid density. Because the pressure is controlled by the
PID, there is no significant change in the pressure over a collection period. Therefore, fluid
density change in the connecting volume due to temperature change between the start and end
of a collection is the only contributor to connecting volume mass storage ( CV ).  The mass flow
at the MUT is given by:  

final initial
MUT LFS CV LFS ∆ cv , (13)

where initial and final are the respective densities in the connecting volume at the beginning and
end of the collection, ∆ is the duration of the collection period, and cv is the size of the
connecting volume.
The size of the connecting volume cv will change with temperature and pressure,
however, it is considered constant in Eqn. 13. This is because changes in cv due to the
pressure and temperature being different than when it was measured (i.e., reference conditions)
are too small to quantify being that a 10 % uncertainty is given to this volume, Section 6.4. By
the same reasoning, any change in cv during a collection due to temperature change is also
negligible.

4. Collecting and Reducing Data

Figure 6 illustrates the reduction technique for the raw data that the 15 kg/s LFS generates.
The traces shown are from the calibration of a 5 cm coriolis meter. The top graphs show the
mass in the collection tank as it is filled. The starting value is the open loop stable mass in the

8 
 
9th ISFFM Arlington, Virginia, April 14 to 17, 2015

tank when the dump valve is initially closed. At the maximum flow of 15 kg/s, generally 50 to 70
data points are collected during a single fill. At the lower flows, the number of data points for a
fill can be as large as several thousand; therefore, the total mass collected can be reduced and
still give enough data points to determine the slope within the desired uncertainty.
Equation 9 is used to yield the mass flow in the collection tank as shown in the middle row of
graphs in Fig. 6. We filter the data to ensure that the slope calculation uses only the portion of
data where the flow is steady. The stability criteria for assessing steady flow conditions
depends on the following: 1) the value of the second derivative of the mass versus time plot
generated from the weigh scale, 2) the standard deviation of flow measurements from the
reference meter, and 3) the maximum and minimum values of the flow measurements from the
reference meter. The bottom row of graphs in Fig. 6 shows the absolute value of the mass flow
rate of change during the collection of mass shown in the top row. Experience has shown that a
data collection is sufficiently stable if after filtering the data the uncertainty in the slope
calculation is less than 0.009 %.

5. Comparison to NISTs Existing LFSs

The 15 kg/s LFS was validated using two of NIST’s existing liquid flow standards. These
consisted of the 65 kg/s LFS [5], which is a larger dynamic, gravimetric standard, and the
smaller 2.5 L/s LFS [6], a piston prover. Figure 1 shows the comparison of calibration data of a
5 cm and a 2.54 cm coriolis mass flow meter. The larger meter was calibrated on the 15 kg/s
LFS and the 65 kg/s LFS. The smaller meter was calibrated on the 15 kg/s LFS and the 2.5 L/s
LFS. The error bars for each LFS is calculated by the expanded uncertainty of that standard
(i.e., 0.021 % for the 15 kg/s LFS, 0.033 % for the 65 kg/s LFS, and 0.064 % for the 2.5 L/s
LFS) root-sum-squared with the standard deviation of the mean of a minimum of three repeated
measurements. The 15 kg/s LFS agrees with the existing standards well within their
uncertainties.
Figure 7 shows the degree of equivalence (En numbers) comparison results, which validate
the claimed measurement capabilities of all three of the NIST LFS’s as demonstrated by the
magnitude of En being less than unity. The degree of equivalence is calculated by:

15 kg/s LFS STD

100 1
MUT, 15 kg/s MUT, STD
n , (14)
15 kg/s LFS STD 2 TS

where is the mass flow, the subscript STD denotes either the 65 kg/s or the 2.5 L/s LFS, and
U is the expanded uncertainty of the standards expressed in percent, and TS is the
reproducibility of the respective transfer standard. Both the 5 cm and the 2.5 cm transfer
standards had a reproducibility of TS = 0.021 %, which is calculated by the standard deviation
of repeated measurements made on the 15 kg/s LFS before and after the comparison.

9 
 
9th ISFFM Arlington, Virginia, April 14 to 17, 2015

Fig. 6. Illustration of data selection.

10 
 
9th ISFFM Arlington, Virginia, April 14 to 17, 2015

Fig. 7. Degree of equivalence for two meters used in a comparison between the 15 kg/s LFS
and the 65 kg/s LFS (blue diamonds) and the 2.5 L/s LFS (red triangles). A 5 cm and a 2.5 cm
coriolis meter was used for comparison with the 65 kg/s LFS and the 2.5 L/s LFS, respectively.

6. Uncertainty Analysis of the LFS

The significant factors that contribute to the uncertainty in the mass flow measurement of the
15 kg/s LFS are: 1) the determination of the slope from the scale mass vs. time data, 2) the
buoyancy correction of the scale reading, and 3) mass storage in the connecting volume. The
Monte Carlo method [4] was used to determine the overall uncertainty of the 15 kg/s LFS
because of the complexity in performing an analytical analysis of Eqn. 9 to determine sensitivity
coefficients for each variable. The expanded uncertainty is 0.021 % over the flow range of
15 kg/s to 0.22 kg/s. The following sections explain the contributing components to the mass
flow uncertainty.

6.1 Slope determination

The weigh scale was calibrated over the range of 90 kg to 635 kg in January of 2013 and in
August of 2014. The scale is linear and a single factor is used to correct its readings. The
calibration factor determined in 2014 agreed with that in 2013 within 0.0003 %. Calibration of
the weigh scale used a sequential, incremental loading method based on two 45 kg weights
calibrated at NIST using the 65 kg weight set and 60 kg mass comparator from NIST’s Volume
calibration service [8]. The calibration procedure requires replacing steel weight of the mass
standards with water and then adding mass standards to sequentially progress across the scale
range. This is done because a sufficient number of mass standards are not available to
complete the calibration without partial water fills. This method has been used to calibrate the
weigh scale used in NIST’s Water Flow calibration service via the 65 kg/s LFS, and is well
documented within the Special Publication for that service [8]. A fit of the calibration data from
the weigh scale to a straight line yields fit residuals within 0.00024 %. The resolution of the
scale is 20 g. Therefore, the scale resolution is what determines the uncertainty in a single
scale reading. However, because the resolution introduces noise in the scale readings that are
centered on the mean value and because we are computing a slope from consecutive readings,
the error introduced by the lack of resolution contributes < 10-7 % to the overall uncertainty in
the mass flow measurement.
The calculation of the slope (Eqn. 9) from the scale mass reading versus time data is the
largest contributor to the uncertainty in the 15 kg/s LFS. It contributes more than 93 % to the

11 
 
9th ISFFM Arlington, Virginia, April 14 to 17, 2015

overall measurement uncertainty. The maximum allowed standard error in the slope calculation
for a measurement to be considered for a calibration point is 0.009 %. To calculate the
standard error of the slope, it is necessary to compute the standard error of regression, which is
given by:

∑
reg
(15)
2

and the normalization factor [4] Sxx:

∑
xx . (16)

Combining Eqns. 15 and 16 gives the standard error of the slope :

2 /
reg
. (17)
xx

Equation 15 estimates the standard deviation of the curve fit residuals of the versus data,
and Eqn. 17 gives the standard error of the slope assuming no uncertainty in the ti values.
The size of the boxcar filter (B) on the slope calculation was also tested. For the data
considered here, the size of the filter did not alter the slope and contributes negligible
measurement uncertainty.

6.1.1 Timing interval

The assumption that ti values have no uncertainty is a good approximation. Tests of the
scale oscillator and the data acquisition clock over a period of three days on three separate
occasions showed a maximum deviation of 0.16 ms / hr, or < 0.00005 % standard uncertainty in
the time increment per data point; which accounts for < 0.00002 % of the overall uncertainty of
the standard.

6.2 Buoyancy corrections

The mass readings of the weigh scale are buoyancy corrected as shown in Eqn. 7.
Buoyancy corrections applied to the scale mass readings are necessary for low uncertainty
measurements. These corrections contribute up to 4.4 % to the overall measurement
uncertainty. The density of air is given by [9]:

kg . . /
air m3 6.65287 10 , (18)

where P is the atmospheric pressure in Pa, RH is the relative humidity expressed in percent,
and T is the air temperature in K. The standard uncertainties in P, RH and T are 0.3 %, 10 %,
and 0.07 %, respectively, as determined by control charts generated from years of calibrations
by NIST working standards. This leads to a standard uncertainty of 0.31 % in ρair.
The density of water at 101 kPa is given by [10]:

kg
water m3 1 , (19)

12 
 
9th ISFFM Arlington, Virginia, April 14 to 17, 2015

where T is the water temperature in the collection tank in ◦C, a1 (◦C) = -3.983035 ± 0.00067,
a2 (◦C) = 301.797, a3 (◦C2) = 522528.9, a4 (◦C) = 69.34881, and a5 (kg/m3) = 999.974950 ±
0.00084. The stated standard uncertainties for a1 and a5 are the only significant contributors for
the computed value of pure water. The uncertainty in the tank water temperature is 0.33 %.
This uncertainty comes from the calibration of the sensor (10 mK) and an added uncertainty of
1 K for spatial variance in the collection tank. The 1 K was determined from temperature
measurements made from the sensors in the 15 kg/s LFS and the environmental monitor. An
added standard uncertainty of 0.01% was used for the density of the water used to validate the
15 kg/s LFS because tap water was used. This leads to a standard uncertainty of 0.03 %
in ρwater.

6.3 Mass storage in the connecting volume

The mass flow in the connecting volume between the MUT and the collection tank is
computed continuously during a calibration. The contribution of this calculation to the overall
measurement uncertainty is less than 3 %.
The density of water in the connecting volume is calculated as a function of both temperature
and pressure:

kg
water m3 ref 1 ref ref , (20)

where ρref is the density of water at 294 K and 101 kPa, β is the thermal expansion coefficient,
and κ is the isothermal compressibility factor for water. The standard uncertainty in the
reference density is 0.0083 % [10]. β (0.0002 / ◦C) is calculated from the best fit line to
◦ ◦
calibration data from 19 C to 23 C and has a standard uncertainty of 0.0015 % for pure water.
REFPROP [11] is used to compute κ for pure water. A standard uncertainty of 10 % was given
to both β and κ because the fluid in this work is tap water.
The pressure and temperature measurements in the connecting volume have uncertainties
due to calibration and spatial resolution. The standard uncertainty in the temperature and
pressure due to calibration is 0.01 % and 0.1 %, respectively. The uncertainty in the
temperature and pressure measurements due to spatial resolution is determined by the
difference in readings between the two temperature sensors in the connecting volume and from
the difference of the pressure sensor downstream of the MUT and the atmospheric pressure,
respectively. A 0.1 K difference has been observed between the two temperature sensors and
therefore, this value is taken as the upper limit in a rectangular distribution. The standard
uncertainty in the temperature is therefore 0.06 %. A difference of 445 kPa has been observed
between the pressure at the MUT outlet and the atmospheric pressure. This value is taken as
the upper limit of a rectangular distribution and gives the pressure measurements a standard
uncertainty as high as 58 %. The standard uncertainty in the connecting volume density is
therefore within 0.05 %.
The size of the connecting volume was determined by measuring the length and diameter of
the piping between the MUT and the pipe exit leading to the collection tank (Fig. 2). This
volume is approximately 0.0095 m3 with a standard uncertainty of 10 %. Therefore, the
standard uncertainty in the connecting volume mass storage is within 17 %.

7. Summary
NIST’s 15 kg/s LFS is a fully automated, dynamic system that utilizes a PID control loop for
flow stability. The LFS uses only the dynamic method in order to avoid the expense and

13 
 
9th ISFFM Arlington, Virginia, April 14 to 17, 2015

complexity of a flow diverter. It can measure flows from 0.22 kg/s to 15 kg/s with an expanded
uncertainty of 0.021 %. The standard has been validated using NIST’s existing primary LFSs.
The uncertainty in the mass flow measurement comes primarily from the slope calculation
from the mass vs. time data generated from the weigh scale, accounting for more than 93 % of
the overall uncertainty. The two other significant contributing factors are the buoyancy
correction to the scale readings and the mass storage in the connecting volume between the
MUT and the standard’s collection tank. The buoyancy correction accounts for up to 4.4 % of
the overall uncertainty. The contribution from the connecting volume mass storage to the
overall uncertainty is as much as 3 %.

8. Acknowledgements
The authors would like to acknowledge and thank Dr. Iosif Shinder and Dr. Michael Moldover
for their invaluable experience and guidance in gravimetric liquid flow metrology.

9. References
                                                            
[1] Shinder, I.I. and Moldover, M.R. Dynamic gravimetric standard for liquid flow: model and
measurements. Proceedings of the 7th Annual International Symposium on Fluid Flow
Measurement. 2009; Anchorage, AK.
[2] Shinder, I.I. and Moldover, M.R. Accurately measuring unsteady water flows using a
dynamic standard. Proceedings of the Measurement Science Conference. 2009;
Anaheim, CA.
[3] Shinder, I.I. and Moldover, M.R. Feasibility of an accurate dynamic standard for liquid flow.
Flow Measurement and Instrumentation. 2010; 21: 128 – 133.
[4] Coleman, H. W. and Steele, W. G. Experimentation and Uncertainty Analysis for Engineers.
3rd ed. New York: John Wiley and Sons Inc.; 2009.
[5] Shinder, Iosif I. and Marfenko, Iryna V. NIST Calibration Services for Water Flowmeters.
National Institute of Standards and Technology (U.S.), Special Publication 250-73, 2006.
[6] Pope, Jodie G., Wright, John D., Johnson, Aaron N., and Crowley, Christopher J. Liquid
Flow Meter Calibrations with the 0.1 L/s and the 2.5 L/s Piston Provers. National
Institute of Standards and Technology (U.S.), Special Publication 250-1039r1. 2014.
[7] Weisstein, Eric W. Boxcar Function. MathWorld. Retrieved 13 September 2013.
[8] Bean, Vern E., Espina, Pedro I., Wright, John D., Houser, John F., Sheckels, Sherry D., and
Johnson, Aaron J. NIST Calibration Services for Liquid Volume, National Institute of
Standards and Technology (U.S.), Special Publication 250-72. 2009.
[9] Jaeger, K.B. and Davis, R.S. A Primer for Mass Metrology, National Bureau of Standards
(U.S.), Special Publication 700-1, Industrial Measurement Series. 1987: 79.

14 
 
9th ISFFM Arlington, Virginia, April 14 to 17, 2015

                                                                                                                                                                                                
[10] Tanaka, M., Girard, G., Davis, R., Peuto, A., and Bignell, N. Recommended Table for the
Density of Water Between 0 ◦C and 40 ◦C Based on Recent Experimental Reports.
Metrologia. 2001; 38: 301 – 309.
[11] Lemmon, E. W., Huber, M. L., and McLinden, M. O. NIST Reference Fluid Thermodynamic
and Transport Properties – REFPROP. NIST Standard Reference Database 23, Version
9.0 User’s Guide. U.S. Department of Commerce, Technology Administration, National
Institute of Standards and Technology, Standard Reference Data Program.
Gaithersburg, MD. November 2010.

15