FEATURE FLOW & LEVEL CONTROL A MODEL APPROACH TO MORE EFFICIENT CALIBRATION
bespoke data acquisition system was setup to log over 100 modbus parameters from multiple Coriolis meters installed in the facilities shown in Fig.1. The following process conditions have
been investigated to date: - • Varying fluid temperature. • Varying fluid pressure. • Ambient air temperature. • Cavitation. • Fluid properties. • Meter misalignment. • Zero drift. • Upstream and downstream flow
obstructions. A classification model, based on the
Gordon Lindsay, technical lead at TÜV SÜD National Engineering Laboratory, describes the work being carried out to provide users with a standardised toolset for successful condition-based calibration
S
ignal transmission and data acquisition in the field of flow measurement has
advanced significantly over the past decade. The traditional analogue signals which have been the backbone of SCADA systems in automated facilities are becoming less dominant in new facility builds as well as facility upgrades. Flow meters and secondary instrumentation (e.g. pressure and temperature sensors) now come with the option of providing a digital fieldbus output, giving end-users more flexibility in communication infrastructures, as well as increasing the volume of data that can be obtained from a single device. For example, a Coriolis flow meter can
output additional process values over a fieldbus, which can be used to infer meter performance, as well as providing the standard fluid flow and density readings for which the devices are known. However, despite having access to this
additional level of diagnostic data, many facilities still operate on a time-based calibration schedule (TBC). This means that high operating costs are incurred by removing a meter for calibration and shipping to an accredited calibration laboratory, bi-annually or annually, irrespective of whether the meter is in fact showing signs of deviating from its previous calibration curve. One reason for this is purely down to
end-user confidence – for many, the financial risks associated with potential errors in flow measurement due to misinterpreting diagnostic data are still
too great. Some plant owners say more case studies of successful condition-based calibration (CBC) implementation in similar industrial environments to their own are needed before taking the leap themselves. In a more technology specific example,
most Coriolis meter manufacturers provide software that allow users to interact and log these digital values. However, detailed analysis of the parameters requires a level of familiarity with the specific characteristics of the facility and their effects on the meter and its digital parameters. In addition, the availability and relevance of certain digital parameters may vary between manufacturers of Coriolis meters. If one moves to a completely different flow metering technology such as ultrasonic, then analysis of different digital parameters is required due to the principles of operation. To address this, we are undertaking
experimentation in our flow laboratories, making use of emerging data science and mathematical techniques to establish correlations in the digital values available from Coriolis meters. The objective is to provide end-users with a standardised toolset that can automatically detect when a Coriolis meter is not performing to specification and crucially, the cause of the deviation in performance. Such an undertaking requires a
knowledge base to be constructed from experiments that investigate the individual and combinatory response of these digital values to specific process conditions. A
20 DECEMBER 2019/JANUARY 2020 | PROCESS & CONTROL Fig. 1: TU V SU D National
Engineering Laboratory’s single phase water flow loop
extensive time-series data files logged during experimentation, has been developed by our Digital Services team. The model has already been shown to be capable of detecting correlations between the meter’s digital fieldbus data and the process effects induced during experimentation. In addition, end-users have also
Gordon Lindsay, technical lead at TÜV SÜD National Engineering Laboratory, a provider of technical consultancy, research, testing and programme management services
expressed an interest in gaining access to simple user interfaces and graphical depictions of individual digital process value importance with respect to detecting specific process conditions. To that end, the model has been developed to include this functionality. Currently in its prototype stage, our tool allows for interrogation of the model through generalised and customisable histograms with respect to known facility process conditions that can cause incorrect meter readings. The tool displays, in simple terms, the relevance of each digital value in determining a specific condition, and as a result provides end users with further knowledge and validation as to the potential decisions that may be taken (automated or human-reviewed) based on the CBC model output. More experimentation is planned in the
coming year, the aim of which is to increase the resolution of data from which the model learns, and in doing so continue to build its accuracy. Successful field implementation of a
model such as this would signify a step change with respect to end-user flexibility in condition-based calibration scenarios. Instead of a red, amber, green (RAG) output, users would have access to diagnostic information that indicated a specific problem that requires investigating or intervention. Therefore, fault diagnosis times would be reduced, resulting in less downtime and maintenance, and lowering operating costs.
TÜV SÜD National Engineering Lab
www.tuv-sud.co.uk/nel
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52
