EVALUATION ON THE CONSISTENCY OF CALIBRATION RESULTS BETWEEN REFERENCE STANDARDS OF PNEUMATIC PRESSURE BALANCE BASED ON NON-FULL RANGE CALIBRATION

To provide calibration services for pressure measuring devices, SNSU-BSN has several piston-cylinder standard that may traceable to different National Metrology Institute (NMIs). Non-full range calibration of pressure balance has been performed to evaluate the consistency of calibration results between those standard, especially for establishing self-traceability in the future. In this research, a piston-cylinder unit S/N 1926 with medium pressure range of 1750 kPa, was calibrated with low pressure range S/N 978 of 350 kPa and high pressure range S/N 1054 of 7000 kPa. The calibration was performed with cross-float method to evaluate the effective area of pistoncylinder at null pressure and reference temperature of 200C (A0,20) and distortion coefficient (λ) as the 1926 main parameters. The obtained value, respectively are (1.961 166 × 10 ± 4.4 × 10) m and (-1.67 × 10 ± 9.4 × 10 ) Pa from 978 and (1.961166 × 10 ± 5.1 × 10) m and (-1.58 × 10 ± 8.4 × 10) Pa from 1054. The result of 1926 from both methods shows good conformity with Normalized Error (En) of 0.0007 and 0.069, respectively. Linearity of effective area changes to the pressure is very consistent in both low and high pressure range. Validation results by using PTB-Germany results, shows the relative different for A0 and λ obtained are less than 0,1 × 10 and 6%,respectively. Therefore, the pneumatic pressure balance of SNSU-BSN is traceable, consistent with each other and capable for disseminating the pressure unit along all primary pressure standard owned with high agreement compared to those of other advance NMIs.


INTRODUCTION
Pressure Balance is the primary standard in the field of pressure metrology, which has been used by SNSU-BSN, as well as many other National Metrology Institute, to provide traceability of pressure measurement for calibration laboratories and industries in Indonesia (Ega & Samodro, 2014). Those pressure balance itself are traceable to the SI unit, either by the SNSU-BSN or through other developed NMIs. As for example the mass of the pressure balance has been traceable to the mass standard owned by Mass Laboratory of SNSU-BSN. However, the effective area of several piston-cylinders are still depends on the primary pressure standard of the other developed NMIs, such as PTB-Germany, KRISS-Korea and NMIJ-Japan. Therefore, a five years period of Pressure Laboratory road map program has been proposed, one of them is to develop the capability in maintaining of primary pressure standard through the independent calibration chain, with means of calibrating its primary pressure standard by the lab itself (Samodro & Ega, 2016), (Samodro et all, 2012).
Typically, the calibration of pressure balance is performed with the full range calibration by using an appropriate or the best matches with the desired pressure range and accuracy of piston gauge standards (Olson, 2009). The full range calibration of pressure balance means that the direct comparison method between pressure balance to be calibrated (DUT) and the reference pressure balance (STD) is performed until the maximum pressure range of the DUT.
However, as the masses are fixed, the pressure generated by the pressure balance depends on the size of its piston-cylinder effective area, which resulting in the different pressure range and limitation of maximum pressure of each piston-cylinder pressure balance can be generated. Meanwhile, the traceability chain is necessary to be realized from low pressure until high pressure (Bair, 2011) (Owen, 2011). Therefore, a non-full range calibration method has been proposed to realize the independent calibration chain. The non-full range calibration of pressure balance means that the direct comparison method between pressure balance to be calibrated (DUT) and the reference pressure balance (STD) is performed until the maximum pressure range of the STD, due to the different pressure range between STD and DUT.
A preliminary study of non-full range calibration method in the calibration of pressure balance has been performed before (Samodro & Ega, 2016). The results was satisfactory proven by the deviation of the effective area A(0,20) less than 2 parts per million (ppm) from the last certificate calibration with full range calibration. Therefore, the research is continued with the purpose to ensure that the reference standards of pneumatic pressure balance owned by SNSU-BSN are traceable and consistent with each other, before conducting the development of selftraceability in pneumatic pressure of SNSU-BSN, by means of self-disseminating of all primary pneumatic pressure standard in SNSU-BSN that has different pressure range with each other.

BASIC THEORY
Pressure Balance is the primary standard in the pressure measurement, which its pressure (P) is defined by weight set that generate force acting on the piston-cylinder (P/C) effective area according to the Equation (1) : (Ginanjar, Ega & Samodro, 2017) ……………..…(1) Where: m = Total loaded true mass on the pistoncylinder assembly, kg g = Local gravity acceleration, m/s 2 Ap,t = Effective area of the piston-cylinder at certain pressure (p) and temperature (t) Typically, the piston-cylinder effective area changes with respect to the pressure (Ap) has the characteristic as linear function of pressure, as shown in Equation (2)  ..…. (5) …………..…. (6) where : = Generated pressure on each pressure calibration point (Pa) = Effective area of PC at each pressure calibration point (m 2 ) = Slope from the linear function curve The type-A uncertainty of each dan that can be determined by calculating standard deviation of Ap, standard deviation of the intercept (A0), standard deviation of the slope ( ) from number of n measurement data according to the Equation (7) until Equation (11) (Morrison, 2014).
………….. (7) ………….. (8) ………….. (9) ………….. (10) .…………. (11) The type-B uncertainty of A0 comes from the effective area of P/C that has been corrected to the reference temperature of 20°C at applied pressure equation, as described in Equation (12) Where: Mi = Total loaded true mass on the pistoncylinder assembly, kg M = Additional trim mass, kg = Air density, kg/m 3 = Mass density, kg/m 3 = Volume buoyancy of the pistoncylinder assembly, m 3 G = Local gravity acceleration, m/s 2 = Applied pressure standard, Pa = Thermal expansion coefficient of the piston-cylinder assembly, °C -1 = Temperature of the piston-cylinder assembly, °C The type-B uncertainty of comes from the slope of linear regression, as shown in Equation (15) 17) Therefore, the combined uncertainty for the and λ are : The conformity of the measurement results obtained from two different methods can be clarified by using the Normalized Error (En) equation, as shown in Equation (20) : …… (20) Where:

XA
= the measurement value from method A XB = the measurement value from method B U(XA) = the expanded uncertainty from method A U(XB) = the expanded uncertainty from method B Both methods results are said to be conformed with each other if the (Ega & Samodro, 2014). This En can also be used to compare both calibration results, from method A and method B, with other value assumed as reference for the validation process.  The calibration was performed at 10 pressure intersection points (80 kPa, 110 kPa, 140 kPa, 170 kPa, 200 kPa, 230 kPa, 260 kPa, 290 kPa, 320 kPa, and 350 kPa) with three measurement series that consists of two increasing pressure measurement series and one decreasing pressure measurement series to evaluate the uncertainty from measurement repeatability.

METHOD
In the second step, the 1926 PCA is calibrated against high pressure range reference standard PCA 1054 up to 1750 kPa. The 1054 is installed on the primary pressure balance with base number 169 with AMH, while the 1926 is installed on the secondary pressure balance with base number 1083 with manual mass load. The calibration is performed at 8 pressure intersection points (525 kPa, 700 kPa, 875 kPa, 1050 kPa, 1225 kPa, 1400 kPa, 1575 kPa, 1750 kPa) with three measurement series that consists of two increasing pressure measurement series and one decreasing pressure measurement series to evaluate the uncertainty from measurement repeatability. The main parameters of the 1926 as the calibrated pressure balance (A0,20 and λ) are calculated by using Equation (4) -(6), while for the uncertainties of both parameters are calculated by using Equation (7) -(19). The conformity of the calibration results between both measurements method are evaluated with En by using Equation (20). The linearity of the calibration has been investigated using all result from both reference standards along the coverage range. Validation of the non-full range calibration results in this research is performed by comparing the main parameters value, as well as its expanded uncertainty obtained between both methods with the value provided from the PTB calibration certificate, that used the fullrange calibration method in their calibration method with direct comparison by means of the cross-float method (PTB, 2017). Figure 5 represents the 1926 PCA effective area changes with respect to the pressure, when calibrated with the 978 and 1054 as the reference standard, respectively. From three series of measurement with two series of increasing pressure and one series of decreasing pressure, it can be seen that the effective area of 1926 decreases linearly as the applied pressure increase. The effective area changes are fluctuates at each pressure point and measurement series, but resulting in linear function of pressure from the average of three measurement series. Figure 4 Calibration results of 1926 with the 978 as the reference standard.

Figure 4 and
As shown in Figure 4 and Figure 5, the standard deviation of the 1926 when calibrated with the 1054 are slightly larger compared to those when calibrated with the 978. The maximum relative standard deviation is 1,9 × 10 -6 from calibration with the 1054, while from the 978 is 0,4 × 10 -6 . It might due to practical reason that the sensitivity of cross-float is much higher using smaller diameter of PCA in relatively high pressure, while gas as the pressure medium is compressible. Figure 5 Calibration results of 1926 with 1054 as the reference standard.. Figure 6 shows the linearity of the effective area changes with respect to the pressure, Ap in the whole pressure range, combination of the calibration results of 1926 from the 978 at low pressure and 1054 at low pressure from the average of three measurement series. The linear curve of 1926 effective area during calibration against 978 up to 350 kPa shows conformity with those of calibrated against 1054 up to 1750 kPa. Therefore, it can be said that the full range of the 1926 maximum capacity, which is 1750 kPa, has been calibrated from 80 kPa until 1750 kPa with high agreement through double calibration by using multiple piston gauge standards.  Table 1, with the uncertainty component described in Table 2. The standard or combined uncertainty of the 1926 effective area at null pressure (A0) with 1054 as the reference standard is larger than those of effective area with 978 as the reference standard, where the expanded uncertainty of A0 with the 1054 is 26 × 10 -6 while with the 978 is 22 × 10 -6 . The main contribution for the uncertainty comes from the Type-B uncertainty of those reference standards 978 and 1054 itself with standard uncertainty of 11 × 10 -6 and 13 × 10 -6 , respectively. While the Type-A uncertainty are small with value of 0.6 × 10 -6 and 2.5 × 10 -6 , for 978 and 1054 respectively as presented in Table 2. It shows that the A0 can be determined accurately with better uncertainty in low pressure, than those of in the higher pressure. λ (Pa -1 ) -1.67 × 10 -12 9.4 × 10 -13 -1.58 × 10 -12 8.4 × 10 -13

ppm
In the other hand, the standard uncertainty of the 1926 distortion coefficient (λ) from both reference are similar and big as presented in Table 3. This due to the some measurement point has poor repeatability due to cross float sensitivity which is illustrated with wide error bar in Figure 6. This could be improved by using another reliable calibration method that could overcome cross float sensitivity method is transducer assisted crossfloat (TAC) method . Combined uncertainty uc(A0) at k = 1 11 13 Expanded uncertainty U(A0) at k = 2,in ppm 22 26 Expanded uncertainty U(A0) at k = 2,in m 2 4.4 × 10 -9 5.1 × 10 -9

Parameter Normalized Error (En)
Ao (mm 2 ) 0.0007 Finally, for the validation, the main parameters of the 1926 values obtained are approximately the same when compared with the value from the calibration certificate provided by the PTB-Germany for the validation, with the value of A0 and λ respectively are (1.961 166 × 10 -4 ± 1.6 × 10 -9 ) m 2 and (-1.67 × 10 -12 ± 1.7 × 10 -13 ) Pa -1 . The relative differences between the obtained A0 and λ value from both methods with the PTB-Germany results are less than 0.1 × 10 -6 and 6%, respectively, with En less than 0.07.
The calibration results of the 1926 from the SNSU-BSN shows very good conformity with PTB-Germany calibration results. Therefore, it can be concluded that the reference standard of pneumatic pressure balance which owned by SNSU-BSN are consistent with each other, according from the calibration results.

CONCLUSION
Experiments and evaluation of the non-full range calibration of pneumatic pressure balance in SNSU-BSN were successfully performed. The calibration results of the 1926 shows good agreement when calibrated against 978 and 1054 with En far less than 1, respectively. This proves that those reference standards are consistent with each other, regardless the differences of the pressure range between them. Moreover, this has been validated by comparing the obtained values of A0,20 and λ with the calibration certificate given by PTB-Germany in which is known as the advance NMI. The relative different between them were less than 0.1 × 10 -6 and 6%, respectively, considering the typical uncertainty for this test PCA is 5 × 10 -6 and 10%.
In addition, the establishment of selftraceability chain, from low pressure range of primary pneumatic PCA standard to high pressure range PCA standard in SNSU-BSN is potentially to be performed by using the proposed non-full range calibration method.