Traceable calibration of automatic weighing instruments in dynamic operation

The article aims to present progress in the development of methods for calibration of automatic weighing instruments (AWIs), which operates in dynamic mode. The specificities of the calibration of the AWIs in the dynamic mode of operation are highlighted. For automatic catchweighing instruments and automatic gravimetric filling instruments, basic information about the measurement method, error model and measurement uncertainty contributions are given. Several approaches are presented for determining the reference mass of the test loads on the control weighing instrument, including the related analysis of the air buoyancy correction and influence of the density of the test loads. The use of harmonized and validated methods will, when finalised, enable a traceable calibration of the AWIs in a dynamic mode of operation.


Introduction
Automatic weighing instruments (AWIs), which perform measurements in a dynamic mode, are extensively used in the preparation, production and quality assurance of pre-packed products as well as for products, whose content or composition is determined by weighing.
While NAWIs are routinely calibrated by accredited calibration laboratories according to EURAMET Calibration Guide for NAWIs [1], the calibration of AWIs is not as well defined, because there is a significant difference between the static measurement mode of operation of NAWIs and the dynamic measurement mode of operation, which is typical for the majority of AWI applications. The growing dissemination of AWIs emphasises the need to confirm their metrological quality by calibrations and the reliable estimation of their measurement uncertainty in order to judge the accuracy of the weighing result. Thus, the development of calibration methods for dynamic measurements with AWIs is the scope of the EMPIR project 14RPT02 "Traceable calibration of dynamic weighing instruments operating in the dynamic mode", with a short name AWICal, funded by EURAMET [2]. This project will develop harmonised calibration methods and uncertainty evaluation models for the three selected categories of AWIs, i.e. automatic catchweighers, automatic instruments for weighing road vehicles in motion and automatic gravimetric filling instruments, selected to represent the most commonly used instruments.

Calibration method
The first step in the process is the establishment of the measurement model, which is a mathematical relation between the measurand and all quantities, which are involved in the measurement. The actual value of weighing is affected by various dynamic effects of instrument operation, the load type and its properties, and the rate of operation of the instrument. Due to the dynamic behaviour of their operation, functional relationship between weighing result, parameters of operations such as rate of operation, type of material and nominal value of the test load is very complex and currently out of the scope of the calibration procedure.
Because of the dynamic behaviour, it will usually not be possible to calibrate a measurement range for such instruments. The calibration will deliver a measurement error of the instrument in comparison to a reference mass value under the conditions of the calibration such as speed of the load transport system and the load properties.
In agreement with the client, the calibration is performed at individual nominal values, which are defined by mass of the test loads, and it is only valid for the specified test loads (with a small bandwidth for mass, volume and shape).
Calibration is performed at the location where the instrument is being installed. Test loads are made of the type of product, which is normally weighed on the calibrated instrument. A calibration is performed at the rate of operation requested and specified in advance by the client. Normally these conditions are the same as conditions during the actual weighing process.
The documents for the calibration of automatic instruments should include at least guidance on: determination the mass of test loads, measurements to be performed, calculation of the measuring results, evaluation of the uncertainty of measurement, contents of the calibration certificates.
In the following two subsections basic details are given about calibration procedures for the automatic catchweighing instruments and the automatic gravimetric filling instruments.

Automatic catchweighing instruments
The object of calibration for the automatic catchweighing instrument, which is symbolically shown on Figure 1, is the indication of the instrument ‫ܫ‬ in response to an applied load with reference value of mass of the test load, ݉ . The error of indication ‫ܧ‬ is Due to effects of air buoyancy and other effects such as convection, drift and others which may lead to minor correction terms δm ୶ , m ୰ୣ is not exactly equal to m ୡ , the conventional mass value of the load ݉ = ݉ ் + ߜ݉ + ߜ݉ … (2) ߜ݉ is air buoyancy correction for the test load used for calibration, and ߜ݉ further corrections that it may be necessary to apply under special conditions. Tests are performed to determine the errors of indications, repeatability of indications, and the effect of eccentric application of a load on the indication.

Test for errors and repeatability of indication
The test consists of the passing repeatedly the same load over the middle of the load receptor, under identical conditions of handling the load and the instrument, and under constant test conditions. The purpose of this test is an appraisal of the accuracy and repeatability of the instrument at least in points of usual application of the instrument. Each test point is characterized by its own repeatability.
For the test load ‫ܮ‬ , the mean error of indication ‫ܧ‬ is calculated as follows where ‫ܫ‬ is the mean of several indications. Section 3 provides further information how ݉ could be obtained. From the ݊ indications ‫ܫ‬ for a given test load, the standard deviation ‫ܫ(ݏ‬ ) is calculated (5)

Test for effect of eccentric loading
The effect of the eccentric application of the load on the indication is tested when applicable. The effect of eccentric loading shall be determined using the selected test load ‫ܮ‬ using the middle of load transport system, the portion of the load transport system that is halfway between the centre and the back, and repeated with the same test load using the portion of the load transport system that is halfway between the centre and the front For each tested band k of the load transport system, the average difference ∆I ୣୡୡ,୩ is calculated as follows:

Standard uncertainty of the error of indication
To account for sources of variability of the indication, i.e. the effect of the resolution of indication, the repeatability of the instrument and the error due to offcentre position of the test load, ‫ܫ‬ ̅ is amended by correction terms as follows The standard uncertainty of the indication of the catchweigher ‫ݑ‬൫‫ܫ‬൯ then equals where ݀ is resolution of the instrument. The standard uncertainty of the indication and the standard uncertainty the reference value of mass ‫ݑ‬൫݉ ൯ contribute to the standard uncertainty of the error of indication ‫)ܧ(ݑ‬ Section 3 provides further information about the standard uncertainty the reference value of mass.

Automatic gravimetric filling instruments
The object of the calibration the automatic gravimetric filling instruments, which is symbolically shown on Figure 2, is the preset value error of the instrument, ‫ܧ‬ determined at different preset values, ‫ܨ‬ , i.e. values, expressed in units of mass, preset by the operator, in order to define the nominal value of the fills. The preset value error for the ݅-th test fill equals: Due to effects of air buoyancy and other effects, which may lead to minor correction terms ߜ݉ ௫ , the reference values of mass of test fill, ݉ , is not exactly equal to ݉ ி, , the conventional mass the test fill. The conventional mass of the test fill cannot be determined directly, but trough determination of conventional mass of the filled container ݉ (gross value) and conventional mass of the empty container ݉ ் (tare value) ߜ݉ is air buoyancy correction for the test fill (refers to net value of test fill), and ߜ݉ further corrections that it may be necessary to apply under special conditions.

Test for preset value error and repeatability
The test consists of repeated filling the fills with the same nominal value, and under constant test conditions. The purpose of this test is an evaluation of the accuracy and repeatability of the instrument at least in points of usual application of the instrument. Each test point is characterized by its own repeatability.
For the preset value ‫ܨ‬ ,, the preset value error ‫ܧ‬ is calculated based on measurement of ݊ test fills as follows When the test fills are weighed on calibrated control instrument (cf. section 3), (12) can be modified by introduction of the mass of a single fill ‫ܨ‬ in the equation, which is determined based on the weighing results for filled and empty containers, W and W , respecitvely. Finally where F is average mass of fills and δm ୭୲ is the total contribution of the air buoyancy correction. Section 3 provides more information how δm ୭୲ is obtained. From the ݊ fills ‫ܨ‬ for a given preset value, the standard deviation ‫)ܨ(ݏ‬ is calculated

Standard uncertainty of the preset value error
To account for sources of variability of the preset value error, i.e. the repeatability of the instrument and the air buoyancy correction, (14) is amended by correction terms as follows The standard uncertainty of the preset error ‫ܧ(ݑ‬ ) equals where ‫ܹ(ݑ‬ ) and ‫ܹ(ݑ‬ ் ) account for the standard uncertainty of the weighing result for filled and empty container, respectively, ‫݉ߜ(ݑ‬ ்௧ ) for the air buoyancy correction, and u൫δF ୰ୣ୮ ൯ accounts for the repeatability of test fills, which is estimated as Further information about the standard uncertainty the air buoyancy correction is provided in Section 3.

Test loads, control instrument and air buoyancy correction
Calibration of AWIs in a dynamic mode of operation can not be performed directly with standard weights. To determine the errors of an instrument, test loads are applied. The test loads should preferably be the type of articles, which are normally weighed on the calibrated instrument. For the purpose of calibration, the traceability of test loads to the SI unit of mass shall be demonstrated using a control weighing instrument. The control instrument may be either separate (a weighing instrument other than the instrument being calibrated) or integral (when a static weighing mode is provided by the instrument being calibrated). The control instrument should ensure the determination of the mass of each test load to accuracy, which is appropriate to the expected uncertainty of calibration of the calibrated instrument. The process of determination of the reference mass value of the test load and its associated measurement uncertainty is one part of the whole process of AWI calibration. An appropriate method of determining the test load reference mass value needs to be selected in a view of required measurement uncertainty and complexity of the method. Some of possible approaches are summarised below: A. Test load is calibrated prior to calibration of AWI. Its conventional mass is reported together with uncertainty, e.g. in a calibration certificate. B. Mass of test load is determined by comparison with standard weight at the time and place of calibration of AWI on a weighing instrument, which is used as a mass comparator. C. Mass of test load is determined by weighing on control weighing instrument at the time and place of calibration of AWI and the control instrument is calibrated at the same time as the calibrated AWI, too. D. Mass of test load is determined at the time and place of calibration of AWI by weighing on control weighing instrument, which had been calibrated prior to calibration of AWI. Results of calibration the control instrument had been reported in the certificate. E. Mass of test load is determined at the time and place of calibration of AWI by weighing on the calibrated control instrument, which satisfy a given specification, e.g. tolerances of [3].
The conventional value of mass of the test load ݉ ‫ܮܶܿ‬ is a priori not known, its density ߩ ‫ܮܶ‬ normally significantly differs from the conventional value 8000 ݇݃/݉ ଷ and the air density ߩ ܽ at the time of calibration is normally different from ߩ 0 (1,2 ݇݃/݉ ଷ ). The reference value of mass of the test load ݉ therefore depends on the air buoyancy correction ߜ݉ and ߜ݉ ்௧ , respectively where ܹ ூ is the result of weighing of the test load on the control instrument and ߩ ௦ and ߩ ூ is the density of standard weights used for adjustment of the control and calibrated instrument, respectively.
It can be seen that the correction is very small and that usually it is not necessary to apply the correction, i.e. ߜ݉ ்௧ = 0. If conformity of the standard weights used for adjustment of the AWI and calibration of the test load to [4] is established and if the AWI is not adjusted before the calibration, the relative uncertainty is evaluated as [1] ‫ݑ‬ Case A could be applicable only in a case when comparable dynamic behaviour of pre-prepared test loads with loads normally weighed on the instruments is achieved. In this case a drift mass of the test load within a time is relevant. A density of the test load needs to be known. Since it is expected that the density considerably differs from 8000 ݇݃/݉ ଷ , air buoyancy correction in not negligible.
In Cases B to E the mass of test loads is determined at the time and place of calibration of AWI. Consequently no significant drift of the test load is expected. A density of the test load does not need to be known. The air buoyancy correction is very small or negligible and does not depend on a density of the test load.
A critical point of Cases D and E is evaluation of uncertainty of the control instrument in use.
In general, if the same control instrument is used, uncertainty of the reference mass value increases from Case B to Case E.

Validation
Finally, it is necessary to experimentally and on-site validate the measurement methods, error models and measurement uncertainty budgets developed and also check the reproducibility of the draft calibration methods and uncertainty budgets developed for the calibration of the selected categories of AWIs with the interlaboratory comparisons.
For this purpose the validation plans were prepared for each group of AWI concerned. The plans take into account instrument selection, test loads selection, determination of the test load mass, preparation and execution of measurements, monitoring environmental condition, calculation of measurement results and reporting.
The results obtained using the calibration methods developed for automatic instruments operating in the dynamic mode will be compared to the reference values obtained by static measurements of test loads on nonautomatic weighing instrument and to the measurements obtained using the automatic instrument calibrated in the static mode of operation, each time taking into account the uncertainty of measurement.

Conclusion
The expected impacts of the EMPIR AWICal project are to provide basis for traceable dynamic measurements on three groups of AWIs based on the draft guides submitted to EURAMET and improved weighing process control in different industries such as production pre-packed products and transport. Consequently, the calibration laboratories for the calibration of AWIs could be accredited by accreditation bodies based on the harmonised guidance. Supporting information on measurement uncertainty will be available to conformity assessment bodies for AWIs.
For AWIs the uncertainty of measurement depends significantly on properties of the calibrated instrument itself, the characteristics of the test loads, the control instrument, and not only on other capabilities of the calibration laboratory. The reference mass of the test load is usually determined on situ. Several methods for determination of the reference mass are available however, it is necessary to look for a suitable compromise between the accuracy and simplicity of the selected method. Additionally, a density of the test load does not need to be known without significant influence to the measurement uncertainty.
Finally, the capabilities developed in emerging EURAMET members during the project will support their local industry and weighing sector and developed conformity assessment competence of these members in order to support the implementation of the MID and the Pre-packages Directives.
This work was part-funded through the European Metrology Programme for Innovation and Research (EMPIR) Project 14RPT02 (AWICal). The EMPIR is jointly funded by the EMPIR participating countries within EURAMET and the European Union.