Project Mustang Part 3: What Would Actually Reach Nearby Homes?

TAYLOR, Texas — Knowing how much pollution leaves a power plant is only half the question for the people who live around it.

The other half is more personal: what concentration could actually reach a nearby home?

Parts 1 and 2 of this WLP investigation looked at the residential geography around Project Mustang and the pollutants associated with the kind of natural-gas and diesel equipment being discussed for the project.

August 2026 Project Mustang materials reviewed by WLP identify a concept involving 84 to 112 natural-gas reciprocating engines rated at 12.5 MW each, potentially producing roughly 1,050 to 1,400 MW, with approximately one exhaust stack per engine. Stacks could reach about 125 feet. Diesel backup and black-start generation have also been disclosed.

Those details tell us what sources need to be examined. They still do not tell a homeowner what would be in the air at a particular address.

That requires air-dispersion modeling.

Emissions and exposure are different numbers

An emissions inventory tells regulators what leaves a source — for example, pounds per hour of nitrogen oxides or carbon monoxide.

That number matters, but it is not the concentration in the air at a house several hundred yards away.

Once exhaust leaves a stack, it rises, cools, mixes and moves. Its path depends on the stack’s location and height, the temperature and velocity of the exhaust, weather conditions, terrain and nearby buildings.

TCEQ describes air-dispersion models as tools used to estimate ground-level concentrations using those kinds of source and atmospheric inputs. Texas modeling guidance calls for source characterization, meteorological data, receptor design and building-downwash information where applicable.

For Project Mustang, that is the analysis that eventually matters most.

A 125-foot stack does not answer the question

Stack height is only one input.

EPA’s AERMOD model also uses factors such as the pollutant emission rate, stack diameter, exhaust temperature, exhaust velocity and the exact geographic location of the source.

A hot, fast-moving plume behaves differently from a cooler plume leaving at lower velocity. Moving a generation block several hundred feet within a 665-acre site can also change the modeled result at neighboring properties.

A taller stack may reduce concentrations immediately beside a source because the plume begins higher in the atmosphere. But the pollution does not disappear. The plume eventually mixes downward, and the highest ground-level concentration can occur farther from the stack.

There is no responsible way to choose that distance for Mustang by looking at a map. It has to be calculated.

The buildings matter too

Project Mustang would not be a collection of isolated stacks in an empty field. Conceptual plans show large data-center buildings, generation areas, substations and other structures.

When wind moves around a large building, it can create a turbulent wake on the downwind side. A nearby exhaust plume can be caught in that disturbed airflow and pulled toward the ground sooner than it otherwise would.

That is known as building downwash.

TCEQ’s modeling guidance specifically addresses when structures are close enough to a stack to require downwash analysis. EPA’s AERMOD system likewise incorporates building-wake effects.

With dozens of possible stacks located around large industrial buildings, downwash is not a minor detail for Mustang. Final building dimensions and stack coordinates could materially change the modeled results.

One prevailing-wind arrow is not enough

Residents naturally want to know which direction the pollution would travel.

The problem is that Central Texas weather changes.

Southerly and southeasterly winds are common, but cold fronts can bring strong winds from the north. Wind speed and atmospheric stability change as well. Some conditions disperse a plume quickly; others can keep pollutants closer to the ground.

That is why a regulatory analysis does not simply draw an arrow on a map and declare one neighborhood “downwind.”

TCEQ updated its preprocessed AERMOD meteorological datasets in March 2026 and provides both one-year and five-year datasets for Texas counties. A refined Mustang analysis would test thousands of hours of weather conditions and identify the combinations that produce the highest modeled concentrations.

The area south of FM 112 could be downwind during one weather pattern. Properties on another side of the project could be downwind during another.

Put the actual homes into the model

Dispersion models calculate concentrations at locations known as receptors.

TCEQ guidance calls for receptor grids dense enough to identify maximum predicted ground-level concentrations and says modelers should consider the location of non-industrial receptors.

For Mustang, a general receptor grid makes sense. But WLP believes the analysis should also include the actual occupied residences around the project.

Those homes do not have to be publicly identified by owner. They could be labeled R-001, R-002, R-003 and so on.

Then residents could eventually see something far more useful than a campus-wide emissions total: the model’s predicted concentration at each receptor under the operating and weather conditions that produce the highest impact.

One engine is not Project Mustang

Scale creates another issue.

A model of one representative engine would not describe a campus containing potentially 84 to 112 engines and stacks.

The analysis should reflect the sources operating together under the scenarios allowed by the final permit — including the main natural-gas engines, applicable diesel testing, black-start equipment and other significant combustion sources.

Startup and shutdown also deserve attention if those periods produce different short-term emission rates. A standby generator may contribute little to annual emissions while still mattering to a short-term concentration analysis if several units operate at once.

Nearby residents will experience the combined site, not one engine at a time.

What AERMOD can — and cannot — tell residents

AERMOD is a regulatory model, not a prediction of what one person will inhale every minute for decades.

It estimates outdoor ambient concentrations based on engineering inputs and historical meteorological data. That makes it useful for testing compliance with air-quality standards and for identifying where a facility’s modeled impacts are highest.

But the result depends on the information entered. Change the stack location, engine model, control efficiency or building dimensions and the answer can change too.

That is why the underlying modeling record matters. The useful questions are straightforward: Which sources were modeled? What emissions rates were used? Which weather dataset was selected? Which buildings were included in the downwash analysis? Where were the receptors placed? What operating scenario produced the maximum result? And where did that maximum occur?

Compliance and zero risk are not the same claim

There is one more distinction worth establishing before the modeling exists.

If a future analysis shows Project Mustang complies with applicable Texas and federal air-quality requirements, that would be important and WLP should report it plainly.

But compliance would mean the modeled project met the standards and permitting criteria being evaluated. It would not mean “zero pollution.” It would not automatically establish “zero health risk.”

The reverse is also true. The presence of pollutants with known health effects does not establish that Mustang will create an unacceptable health risk at nearby homes. Concentration and dose matter.

That is why the model matters.

The next meaningful document

The next major step in this investigation will come when project-specific modeling becomes available.

The most useful public record will not be another general assurance that the project will be clean or compliant. It will be the engineering record showing where the sources are, where the receptors are, what assumptions went into the model and what concentration is predicted at the places people actually live.

Until then, the conclusion has to remain limited.

Project Mustang has enough proposed combustion equipment to justify detailed residential exposure modeling. The public record does not yet tell us what that modeling will show.

Part 4 will move from modeling to real-world experience, examining what residents, regulators and courts have documented around other large private-power data-center campuses — and which comparisons are genuinely relevant to Taylor.

Sources

• Project Mustang materials reviewed by WLP — August 2026.
• Texas Commission on Environmental Quality — Air Quality Modeling Guidelines
• TCEQ — AERMOD Meteorological Data
• U.S. EPA — AERMOD / Preferred and Recommended Models
• Wikimedia Commons — Moss Landing Power Plant photograph, Thomson200, CC0