← All case studies
Case 03Water & Infrastructure

The warehouse under the city’s utility reporting

ETL pipelines and Kimball-style dimensional models in a Databricks medallion architecture, establishing the data foundation for predictive maintenance, long-term planning and compliance reporting on critical city infrastructure.

Client
Local government
Year
2025
The Problem

Reporting on utility infrastructure had been assembled per question — each analysis pulling from source systems in its own way, with its own assumptions about what a meter, a main or a service connection was.

That works until the questions get long-range. Predictive maintenance and long-term asset planning need a stable dimensional history, and compliance reporting needs the same numbers to come out the same way twice. Neither is achievable against ad hoc extracts.

What We Did
  1. Designed and built ETL pipelines landing enterprise SAP data into a Databricks medallion architecture — bronze for raw fidelity, silver for conformed entities, gold for the reporting layer.
  2. Modelled the gold layer in Kimball-style dimensional form, so facts and conformed dimensions are shared across reporting rather than rebuilt per report.
  3. Integrated the modelled output into the organisation’s reporting platforms as the common source for utility infrastructure analysis.
  4. Reviewed legacy code and system workflows along the way to troubleshoot data issues and reduce the technical debt the old extracts had accumulated.
The Result
  • Data foundation established for predictive maintenance on critical city utility infrastructure.
  • Long-term infrastructure planning and compliance reporting run from the same modelled base.
  • Reporting logic consolidated into conformed dimensions instead of being re-derived per analysis.
Technology
DatabricksMedallion architecturePySparkPythonKimball dimensional modellingSQL ServerMicrosoft Azure

Have something like this sitting on your desk?

If any of the above sounds like a problem you're carrying, a free, no-obligation consultation is the cheapest way to find out whether it's tractable.

Next — Case 04: Taking hundreds of manual tasks out of building consents