Professional Documents
Culture Documents
Epjconf Efm2017 02030
Epjconf Efm2017 02030
1051/ epjconf/201714302030
EFM 2016
Abstract. The article presents estimation of measurement uncertainty of a liquid mass flow using the
orifice plate. This subject is essential because of the widespread use of this type of flow meters, which
renders not only the quantitative estimation but also qualitative results of this type so those measurements
are important. To achieve this goal, the authors of the paper propose to use the theory of uncertainty. The
article shows the analysis of the measurement uncertainty using two methods: one based on the "Guide to
the expression of uncertainty in measurement" (GUM) of the International Organization for Standardization
with the use of the law of propagation of uncertainty, and the second one using the Monte Carlo numerical
method. The paper presents a comparative analysis of the results obtained with both of these methods.
*
Corresponding author: anna.golijanek-jedrzejczyk@pg.gda.pl
© The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution
License 4.0 (http://creativecommons.org/licenses/by/4.0/).
EPJ Web of Conferences 143, 02030 (2017 ) DOI: 10.1051/ epjconf/201714302030
EFM 2016
C S
qm H d 2 2'pU1 (1) 2.1 Analysis of the measurement uncertainty
1 E 4 4 according to GUM
where:
C – a discharge coefficient [-], Complex uncertainty uc(qm) determining the mass
d– orifice diameter [m], flow qm is defined as follows [1]:
E – relation between orifice and pipe diameter:
uc qm u 2A qm u 2B qm (4)
d
E (2)
D where: u qm - uncertainty Type A, u qm -
A B
uncertainty Type B.
D – pipe diameter [m],
To determine Type A uncertainty, the probability
H - compressibility of different fluids. For water and
distribution of values of the observations was examined.
other liquids H=1. Most frequently normal distribution is assumed 'a priori'
'p – pressure difference [Pa], (especially when the number of measurements is greater
U1 – density of medium [kg/m3]. than 30).
The estimate of the mass flow was determined as [1]:
When substituting equation (2) into (1):
n
1
qm
C
H
S
d 2
2'pU1 (3)
qm
n
¦ qmi (5)
4 4 i 1
§d·
1 ¨ ¸
©D¹ The standard deviation s of an observation was
determined from the dependency:
The mass flow qm is determined indirectly, and the
function of the measurement is dependent on the
following parameters qm = f (C, d, D, 'p, U1). 1 n
s ¦
n 1 i 1
(qmi qm)2 (6)
2
EPJ Web of Conferences 143, 02030 (2017 ) DOI: 10.1051/ epjconf/201714302030
EFM 2016
The standard uncertainty uA(qm) of the measurement Table 1. The weight coefficients arranged to equation (8)
is evaluated as [1]: (partial derivatives).
s Partial
uA (qm ) (7) Formula
derivatives
n
Assuming no correlation between the uncertainties of 1 S
d 2 2'pU 1 (10)
measured quantities, according to the law of the ª kg º
§d· 4
4
1 c1 « »
uncertainty propagation [1], the one of Type B ¬ s ¼ 1¨ ¸
determining the mass flow qm is defined as follows: ©D¹
2d 3
SC 2'pU 1 (11)
ª kg º 3
2 c2 « § § d ·4 · 2
¬ m s »¼ D4 ¨1 ¨ ¸ ¸
c12u 2 C c2 2u 2 d ¨ ©D¹ ¸
© ¹
u B qm c3 2u 2 D c4 2u 2 'p
c5 2u 2 U1 d4
SCd 2 2'pU 1 (12)
ª kg º 3
3 c3 « § § d ·4 · 2
¬ m s »¼
2 2
§ wqm · 2 § wq · 2 2D5 ¨1 ¨ ¸ ¸
¨¨ ¸¸ u C ¨¨ m ¸¸ u d (8) ¨ ©D¹ ¸
© wC ¹ © wd ¹ © ¹
2 2
§ wqm · 2 § wq · S
¨¨ ¸¸ u D ¨¨ m ¸¸ u 2 'p C
d 2 U1 (13)
© wD ¹ © w'p ¹ c4 >m s @ § 4
· 4
2'pU1 ¨¨ 1 §¨ d ·¸
4
2 ¸
§ wqm · 2 ¸
¨ ¸ u U1 © ©D¹ ¹
¨ wU ¸
© 1 ¹
C S
d 2 'p (14)
where: ª m3 º
§ · 4 4
2'pU1 ¨¨ 1 §¨ d ·¸ ¸¸
5 c5 « »
c1 to c5 – sensitivity coefficients, ¬ s ¼
uB(qm) – the estimated uncertainty of measurement of the © ©D¹ ¹
mass flow qm with the Type B method,
u(C) – the uncertainty in the discharge coefficient C,
Consequently the variances of the orifice diameter
u(d) – the uncertainty of orifice diameter measurement,
u2(d) and pipe diameter u2(D) were estimated on the
u(D) – the uncertainty of pipe diameter measurement,
assumption of the rectangular probability distribution of
u('p) – the uncertainty of the differential pressure
the orifice d and pipe D diameters tolerances, as:
measurement,
u(U1) – the uncertainty of the density of medium 2
§ d ·
measurement. u 2 d ¨¨ ¸¸ (15)
In order to estimate the uncertainty of Type B, the © 3¹
equations describing the sensitivity coefficients 2
§ D ·
appearing in equation (8) were established first. u 2 D ¨¨ ¸¸ (16)
Following that the values of sensitivity coefficients are © 3¹
summarized in Table 1 along with their units.
The weight coefficients in equation (8) (partial where 'd and 'D are tolerances of the apropriate
derivatives) were determined and are also included in diameters.
Table 1.
The differential pressure 'p is usually measured by
The next step of the uncertainty analysis was to
differential pressure transducer, which is made with the
estimate the following variances:
accuracy '('p). The differential pressure variance
a) discharge coefficient C – u2(C),
u2('p), resulting from the processing error at the
b) orifice diameter u2(d),
pressure transducer was determined on the assumption of
c) pipe diameter u2(D), the rectangular probability distribution, as:
d) differential pressure u2('p),
e) density of a medium u2(U1).
§ ''p ·
2
The variance of the discharge coefficient u2(C) was u 2 'p ¨¨ ¸¸ (17)
estimated on the assumption of the normal probability © 3 ¹
distribution for estimation of the discharge coefficient C,
as: The following discussion assumed that density of the
2
§ C · medium U1 is constant. The variance of this density
u C
2
¨ ¸ (9)
© 2 ¹ u2(U1) was determined also on the assumption of the
rectangular probability distribution, as:
where 'C is the estimation error of the coefficient C.
3
EPJ Web of Conferences 143, 02030 (2017 ) DOI: 10.1051/ epjconf/201714302030
EFM 2016
§ 'U ·
2 especially effective in case where the analytical
u 2 U1 ¨¨ 1 ¸¸ (18) description of the phenomenon is complicated.
© 3 ¹ This method is a numerical tool which simulates
generally an unlimited number of unique measurements
After determining the combined uncertainty uc(qm) by random sampling from the known probability density
according to formula (4), one should calculate the function of all input quantities and propagates their
expanded uncertainty Up, which is a measure of the distributions for the measurement model as the output.
required quality of mass flow qm. The expanded The Monte Carlo procedure is conducted as follows:
uncertainty Up is determined following the formula: 1) the number M of the trials should be selected,
2) generate M vectors by random sampling from the
U p q m k p u c q m (19) probability density function for the (set of N)
input quantities,
3) for each vector, evaluate the model to give the
The most commonly accepted value of the coverage
corresponding output quantity,
factor kp is equal to 2 for confidence level p = 95%.
4) estimate of the output of the model,
However, it is more accurate to determine the
5) sort the model values into non-decreasing order,
coverage factor kp=tp(Qeff) on the adopted probability
6) use the sorted values to estimate the uncertainty
distribution and the basis of the combined uncertainty
for the output.
values uc(qm), as well as the knowledge concerning the
In metrology it is used to verify the estimates of
number of degrees of freedom for the uncertainty Type
analytical uncertainty, especially in cases when dealing
A - QA and Type B - QB. with an indirect measurement of the measured value [16-
The Qeff signifies effective degrees of freedom and 17] and when the measuring function is non-linear.
can be obtained from the expanded Welch-Satterthwaite
formula [1]:
3 The example
u c4 q m
Q eff (20) In order to better illustrate the methodology
u A4 q m u B4 q m presented above for estimating the uncertainty of
QA QB measurement, it was decided to attach an example of the
mass flow qm measurement. This measurement is carried
The number of degrees of freedom QA was designated out as shown in Fig. 1. The following data was adopted -
as [1]: Table 2 (measurement conditions are compliant with the
ISO Standards [18-19]).
QA n 1 (21) Table 2. Data.
1
QB (22) 2 d [m] 0.073648
2 G rel
2
,B
3 D [m] 0.100051
The values of Qeff were read out from t-Student’s
table as corresponding to coverage factors kp. 4 H [-] 1.000000
Those results should be presented as the mass flow
estimate qm with the expanded uncertainty Up (qm): 5 'p [Pa] 2753.400
6 U1 [kg/m3] 1.109800
qm r U p qm kg
s
(23)
7 E [-] 0.736100
4
EPJ Web of Conferences 143, 02030 (2017 ) DOI: 10.1051/ epjconf/201714302030
EFM 2016
Table 3. Data of the mass flow qm. Table 5 shows the values of the variance for the
assumed probability distributions and the relative errors
i qmi [kg/s] i qmi [kg/s] i qmi [kg/s] G of the various parameters measurements.
5
EPJ Web of Conferences 143, 02030 (2017 ) DOI: 10.1051/ epjconf/201714302030
EFM 2016
6
EPJ Web of Conferences 143, 02030 (2017 ) DOI: 10.1051/ epjconf/201714302030
EFM 2016
Table 7 summarizes the results obtained for Table 7. The expanded standard uncertainty of the mass flow.
estimation of value of mass flow qm and the expanded
uncertainty Up(qm) estimated with both; the method of Mass flow estimation qm
propagation of uncertainties (traditional one), and the
p Traditional method Monte Carlo
Monte Carlo one. kp
It can be observed that the results of measurement [%] 10-3[kg/s] 10-3 [kg/s]
uncertainty of the mass flow by both methods are 95 2.01 (239.57±2.88) (239.54±2.77)
similar. The relative uncertainty is equal to, respectively:
1.20% - by the traditional method, and 1.15% by the
Monte Carlo one.
7
EPJ Web of Conferences 143, 02030 (2017 ) DOI: 10.1051/ epjconf/201714302030
EFM 2016