SUNConferences, 29th Annual Conference of SAIIE

Font Size: 
Waldt Hamer, Walter Booysen, Edward Henry Mathews

Last modified: 2018-10-09


Data quality is fundamental in quantifying and reporting industrial-scale energy consumption. A narrow focus on isolated aspects of data quality can lead to inconsistent levels of rigour in dataset evaluation and selection. In this paper, a data quality framework is therefore developed to holistically evaluate if a dataset provides a fair representation of the underlying energy system.
The developed framework is based on three core aspects of data quality, namely accuracy, integrity and relevance. The framework emphasizes the use of traceability pathways to test data integrity and relevance. Quantitative and qualitative comparisons are proposed as practical options to test and evaluate multiple data sources. Based on the framework, a dataset can either be validated for reporting purposes or discarded based on the lack of data quality assurance.
The framework is applied to six isolated case studies. The results indicate that discrepancies relating to data integrity and relevance can significantly impact reporting functions. If left unchecked, these quantifiable discrepancies could result in data-based errors amounting to R1240 million (at R0.95/kWh) if viewed within the context of the Section 12L energy efficiency tax incentive. This highlights the role of holistic data quality evaluation to avoid propagation of erroneous data into reporting functions.

Full Text: PDF