Urban heat grids need to be resilient against several factors including peak loading, uncertain demand and outages of pipelines or heat sources. We present a practical, efficient and robust method for the cost-optimal planning and analysis of heat grids. The algorithm combines industry guidelines, geometric search and convex optimization to determine the best heat grid topology. Using an example data and assumptions for the city of Zurich, we find that (i) connecting heat grids to large areas is advantageous since it tends to reduce the grid costs, (ii) outage resilience of the backbone grid only gradually increases the grid costs by about 20%, and (iii) bad demand predictions are very disadvantageous since they may easily double or triple the grid costs.

