Sewer managers in Pearland, Texas, had every right to be astonished at the revised projections for their 70-year capital improvement program: A new computer model showed their department could almost halve the total number of lift stations by increasing conveyance capacity, a change that would reduce projected capital costs by $68 million. This change would not only reduce capital costs drastically, it would also cut operating and maintenance expenses for the upgraded system by $400,000 to $500,000 annually.

The reason the new model produces such different results goes back to the data and the modeling system used. The first model, created in 2000 by another consulting firm, developed a capital improvement plan based on a small fraction of the city's manholes, sewers, and lift stations, which were modeled as passive or “pass-through” facilities. Peak flows were based on general assumptions that weren't verified against actual flow during dry and wet weather. Estimates of infiltration and inflow (I&I) were mostly guesses.

In Pearland, it's easy to guess wrong about infrastructure: In this booming Houston-commuter suburb of 87,100 residents, hundreds of new sewer and water connections every month are patched into a hodge-podge system that dates from the 1930s. There are newly developed areas that flow through mid-life areas, which flow through really old areas before they get to a treatment plant.

In some areas, poor planning had allowed wastewater from a new development to flow through older lines never sized for the additional volume. In other areas, wastewater passes through as many as five lift stations before reaching a treatment facility. The operations of all this infrastructure were tracked manually by operators writing in field notebooks and supervisors transcribing the numbers into a central log.

In 2003, the city hired us to create a master plan that would include wastewater, water distribution, and roads, but without changing the ‘guesswork' wastewater model used to develop the first capital plan. While it's relatively easy to look at long-range land use, it's a good deal more complicated to accurately project infrastructure requirements, particularly given the patchwork land use in the city. We made the best recommendations we could based on the data available.

The city asked us to verify our wastewater recommendations using the old model. Digging into this model, we found it woefully inadequate. The city then asked us to build a better model.

In early 2006, we began developing a comprehensive dynamic model using Wallingford Software's Info Works program. This model simulates reality very closely: It includes almost every manhole and piece of property within city limits and its extra-territorial jurisdiction (ETJ), and it models lift stations directly and dynamically—not by guess work and inference. By implementing build-out scenarios within the city and its ETJ in the new model, we could define where the problem areas would lie and how to address problems strategically and cost-effectively.

The trick in making the dynamic model work was, of course, having the right data. In particular, we needed to correlate wastewater flow data to rainfall events. When we asked the city to collect this data, we were pleasantly surprised to discover that more than a year of data had already been collected and was easily available.

In 2005 the public works department had implemented a data collection and reporting system offered by BirdNest Services Inc. to track and manage water and waste water operations. The system collected detailed data on rainfall and daily pump run-times, as well as the pump nameplate data necessary to convert run-times to flow. Here was a mass of complete, consistent, uniform data for the whole wastewater system—a level of quality that we'd never seen before for this type of project—and all readily available in Excel workbook format.

This data allowed us to create a model that would closely reflect reality. The year's worth of run-time data provided a wealth of information about how each lift station reacted to rainfall events. We also collected several years of wastewater discharge reports for city-owned treatment plants so we could balance our numbers and calibrate the model.