Field Performance and GPS-X Simulation Fidelity of a Full-Scale Municipal Wastewater Treatment Plant in Jordan: A Case Study of the Irbid Central Wastewater Treatment Plant (Fuara)

Water is the lifeblood of any city, but the water that leaves our homes and businesses carries a heavy burden of waste. Before this wastewater can be returned to the environment or reused for farming, it must pass through a treatment plant, a complex facility designed to strip away harmful organic matter and solid particles. In many parts of the world, engineers rely on sophisticated computer programs to predict how these plants will behave, helping them decide when to upgrade equipment or how to handle sudden surges of dirty water. These digital models act as virtual laboratories, allowing experts to test changes without risking the actual flow of water. However, for these simulations to be truly useful, they must mirror reality with high precision. If a computer program predicts that a plant is cleaning water better than it actually is, the consequences could be environmental damage or a failure to meet safety standards.

In northern Jordan, where water is a scarce and precious resource, the Irbid Central Wastewater Treatment Plant, known locally as Fuara, serves a large population including the city of Irbid and its surrounding communities. This facility, which has been in operation since the late 1980s, uses a process called extended aeration to break down waste. In this method, air is pumped into large tanks to encourage bacteria to consume organic pollutants, effectively turning dirty water into cleaner effluent that can be used for irrigation. The plant operates under a national standard that sets strict limits on how much pollution can remain in the water before it is released. Because the region faces chronic water shortages, the reliability of this plant is not just a local concern but a matter of national strategy. To ensure the plant continues to perform well, researchers recently spent a year watching it closely and then built a digital twin of the facility to see how well a computer could predict its real-world performance.

The researchers collected samples of the water entering and leaving the plant every week throughout 2022. They measured three key indicators of water quality: the amount of oxygen-hungry organic matter, the total amount of dissolved and suspended chemicals, and the quantity of solid particles floating in the water. The results showed that the physical plant was doing an excellent job. Over the course of the year, the plant successfully removed more than 97 percent of the organic matter and over 96 percent of the solid particles. The water leaving the facility consistently met the strict Jordanian safety requirements, proving that the physical infrastructure was functioning as intended. The team then took this year's worth of real data and fed it into a computer simulation software known as GPS-X, a tool widely used by engineers to model wastewater treatment processes. They adjusted the settings in the model to match the plant's actual performance, hoping to create a reliable digital replica that could be used for future planning.

When the researchers compared the computer's predictions to the actual water samples, the results revealed a surprising split in accuracy. The model was remarkably good at predicting the removal of organic matter, matching the real-world data with a difference of only about six percent. This suggests that the computer understood how the bacteria were eating the waste. However, the model struggled significantly with the other two measurements. When it came to predicting the levels of total chemicals and solid particles in the clean water, the computer was off by large margins, underestimating the amount of pollution by more than 40 percent for chemicals and over 70 percent for solids. In simple terms, the digital model thought the plant was producing much cleaner water than it actually was. The simulation assumed the solids were settling out perfectly, while in reality, a significant amount of sludge was still floating in the final output.

This discrepancy highlights a critical lesson for engineers and plant operators. The study demonstrates that a computer model can be tuned to get one part of the process right without getting the rest right. Just because a simulation accurately predicts how well organic waste is broken down does not mean it can reliably predict how well solid particles are removed. The researchers found that the model's failure to account for the messy, real-world behavior of the settling tanks—where factors like scum formation and sludge depth vary constantly—led to these large errors. The study argues against the common practice of assuming that a model calibrated for one parameter is automatically reliable for all others. Before such digital tools are used to make expensive decisions about expanding a plant or installing new technology, they must be tested against every single type of pollution they are expected to manage. For the Fuara plant and similar facilities, this means that while the computer is a helpful guide for understanding biological processes, it cannot yet be trusted to predict the behavior of solid waste without further, specific adjustments.