Data reconciliation –
Create transparent processes

Data reconciliation
quality control for
the whole process

If the quality criteria of VDI 2048 are not fulfilled, the causes need to be determined and corrected. In order to do this, the system supplies the user with indications about gross measurement errors (suspected tags) which then need to be newly calibrated. If the calibration does not indicate an erroneous measuring device, the estimated 95% confidence interval of the measurement needs to be increased.

The result is a description of the state of the process that is reflecting reality with the highest possible probability and at the same time with the smallest uncertainties. Currently used, highly precise measurement devices such as ultrasonic flow meters, tracer, etc. are singular pieces of information and cannot reach the quality controlled level of precision of PDR.

Process data reconciliation (PDR) as described in the German guideline VDI 2048 is based on the Gaussian compensation theorem and represents the best possible quality control for quasi-stationary flow processes (e.g. in power plants or pipelines). The goal of PDR is to determine the most likely process state with the smallest possible uncertainties. This condition can only be reached when the quality criteria described in the VDI 2048 guideline are met.

MODELING
n a first step, the process is mapped into a thermodynamic model. Secondly, all available process data such as measurements (temperatures, pressures, mass flows, et.) are allocated to the model. At the same time, all process data needs to be provided with a 95 % confidence interval. This model is then solved with the Gaussian compensation theorem under the condition of exactly fulfilling closed mass, energy and chemical balances (terminating condition: 1 x 10-6). After this step all process information will have lost its character as single pieces of information, i.e. measurements are no longer independent parameters.