At 550 horsepower, the stock L5P turbo is already at the edge, and boost by itself doesn’t explain why. The real story is air density. Ambient conditions set the starting point, the turbo adds density by compression, and the intercooler adds more by cooling that charge back down. If you want to know how much horsepower the engine can really support, that’s the number to watch. That also means the intake system matters before the compressor ever does any work. Any pressure drop or temperature gain between the room and the turbo inlet costs density, and once density is gone, the turbo has to work harder to get it back. A large, low-restriction cold-air intake helps preserve that inlet density, while a hot underhood intake gives it away. Using Banks iDash Pro with pressure and temperature sensors across the intake path makes that visible in real time. You can see ambient density, compressor inlet density, pressure drop, and the engine’s airflow demand instead of guessing from boost alone. In this pull, the intake loss was small, but the stock turbo was already out of speed and drive-pressure margin. That’s how you find the real limit without blindly pushing parts until they fail.
In this installment of Killing the Duramax, Gale Banks outlines the strategy for pushing an L5P Duramax to failure in a controlled way. Rather than simply adding power until parts scatter, the plan is to examine each system that contributes to manifold air density, identify the limiting component at each stage, upgrade that component, and continue until the engine is physically broken. The emphasis is not on boost alone, but on the complete chain of airflow, compression, cooling, and cylinder pressure that determines how much power the engine can safely make.
Banks explains that every relevant system has been instrumented with pressure, temperature, or speed sensors, and the resulting data is processed and displayed on the iDash. That allows the team to see exactly how much air density each component contributes and where losses occur. The goal is to show not just what fails, but why it fails, and to do so in the engineering order that the engine experiences it.
The central concept in the episode is that horsepower potential in a diesel is tied to air density, not to boost pressure by itself, and not to temperature or humidity considered in isolation. Banks gives a standard-day reference value of 72.4 pounds of air per 1,000 cubic feet, noting that this is roughly the air density in the room. That amount of air mass is enough to support about 600 horsepower in this diesel.
The complication is that the Duramax cylinders displace only about 350 CFM at 3,000 rpm. To make 600 horsepower, the engine effectively needs the air mass contained in 1,000 cubic feet of ambient air, so that air must be forced into the engine. Ambient pressure plus boost pressure provides the force, but density is the quantity being forced. At the current operating point, a little over 550 horsepower, Banks says roughly one-third of the manifold air density comes from the ambient air in the room, another third is added by the turbocharger, and the final third is contributed by the intercooler. His point is that the intercooler is not merely reducing temperature; it is directly increasing charge density and therefore power potential.
The airflow path begins with what Banks repeatedly describes as a very large filter, the same filter used in a Banks Ram-Air system for the L5P. He says the design objective was to maximize filter size, minimize pressure drop, and maximize dirt capacity rather than accept the compromises common in the aftermarket. From the filter, air passes through a 5-inch mass airflow sensor and then into a 5-inch tube that gradually reduces into the OEM turbo inlet.
The intake tract is instrumented to measure both temperature and pressure entering the compressor. Pressure is routed through a nylon line to a sensor, then through a Banks data module and onto the network feeding the iDash displays in the control room. This setup allows the team to quantify pressure drop and temperature rise across the intake system before the air even reaches the compressor. That is important because any loss at the compressor inlet reduces the density available to the turbocharger and therefore reduces the density that can ultimately reach the cylinders.
Before starting the engine, Banks reviews the ambient-condition channels on the iDash. The ambient air density in the control room is 71.6 pounds per 1,000 cubic feet, which he describes as 99 percent of a standard day. He calls this percentage one of the most important values to monitor because it directly reflects the engine's power potential before any compression or cooling takes place. Based on that 71.6-pound density, the display predicts 598 horsepower for diesel operation if that air mass can be delivered to the cylinders.
The ambient air pressure is shown as 14.4 PSI absolute, or 29.3 inches of mercury. Banks stresses that he is discussing measured air pressure at the location, not corrected barometric pressure of the kind used in weather reports. He frames naturally aspirated breathing in terms of pressure above a near-vacuum created by the descending piston, arguing that pressure, not vacuum, is the useful engineering language. Ambient air temperature is also displayed, along with relative humidity, which is 40.9 percent. He notes that this humidity is reducing density somewhat; if humidity were zero, as assumed in many standard-day definitions, the ambient power potential would be about 2 percent higher.
A second iDash layout shows density altitude and water content. At the time of the test, density altitude is 1,840 feet. Banks notes that racers and pilots commonly use this value because it expresses current atmospheric conditions as an equivalent altitude under standard-day assumptions. The display also shows water content directly as 53.1 grains of water per pound of dry air, which he considers more informative than relative humidity alone.
The instrumentation also calculates engine airflow demand. Because the engine size has been preloaded into the system, the iDash can use rpm to calculate CFM engine, meaning the airflow being pumped by the pistons. The standard-day reference in use is SAE J1349. Another display page shows ambient pressure, ambient air temperature, intake-system pressure drop, and compressor inlet pressure. With the engine off, those values are equal, but under load the team expects to see pressure drop and possibly some temperature gain. In the dyno cell, temperature rise through the intake is minimal because the system is not drawing hot underhood air. Banks contrasts this with vehicle installations that use exposed underhood filters, arguing that ingesting hot engine-bay air directly reduces density and therefore performance.
As the engine warms, Banks shifts attention to a device called the Kistler KiBox, which monitors in-cylinder pressure. He says the nominal design limit for this engine is 180 bar maximum cylinder pressure, and he wants to push beyond that while still understanding exactly what the pressure trace is doing. The KiBox provides not only peak cylinder pressure but also the shape of the pressure curve, including where injection begins, where combustion pressure rises, where peak pressure occurs, and how pressure decays during expansion.
At low load, the displayed trace shows injection beginning before top dead center, which Banks estimates at roughly 7 to 8 degrees BTDC. The pressure peak occurs after top dead center, and at this warm-up condition the engine is showing about 63 bar with engine speed around 1,120 rpm. He emphasizes that this instrument is as important as the dynamometer because cylinder pressure is the physical mechanism that will ultimately kill the engine. The objective is to avoid destroying it prematurely by using rpm, airflow, and other means to increase output before simply driving cylinder pressure high enough to put the crankshaft on the floor.
When the team begins the loaded run, Banks notes an important calibration detail: to make 550 horsepower with the stock variable-geometry turbocharger, they actually have to close the vanes somewhat to speed the turbo up. This is done with a special calibration created specifically to reach that power level. He states that the turbocharger is already at its limit around 515 horsepower, so the 550-horsepower target is beyond the comfortable operating range of the stock unit.
The engine is brought up to 3,000 rpm, where piston demand is about 350 CFM, and then loaded to roughly 550 horsepower. Banks points out that the horsepower shown is a raw reading rather than a corrected value. At this operating point, the intake system itself performs well: pressure loss from the room to the compressor is low, temperature rise is about 1 degree, and the density entering the compressor is still 98.8 percent of ambient. In other words, the cold-air intake is costing almost a full percent of density, mostly due to pressure drop, but very little due to temperature in the dyno-cell environment. He notes that the loss would be greater in a truck because underhood heat would raise inlet temperature.
Although the intake system is behaving well, the turbocharger is not. At the 550-horsepower operating point, cylinder pressure has climbed to about 190 bar or slightly higher, already beyond the engine's nominal 180-bar design pressure. Banks says the turbocharger is "out to lunch," meaning it is operating outside a reasonable efficiency and durability range. Turbine drive pressure is far off target, and turbo speed is around 135,000 rpm, which he identifies as approximately the design-speed limit.
That combination of excessive cylinder pressure, excessive turbo speed, and poor turbine-side conditions tells the team they have reached the practical limit of the stock turbocharger. Rather than continue and risk an uncontrolled failure, they back the engine down. Banks makes clear that this restraint is intentional: the purpose is to identify the limiting component scientifically, not to create internet-style destruction without understanding. In his view, this is exactly the kind of situation where tuners without a dynamometer or proper instrumentation push hardware beyond safe limits and then mistake the result for achievement.
The episode ends without a catastrophic engine failure, but with a clear conclusion about the next bottleneck. The ambient conditions are nearly ideal, the intake system is preserving most of the available density, and the instrumentation is showing exactly where the losses and stresses are occurring. The stock turbocharger has become the limiting component before the engine itself is intentionally broken.
That sets up the next phase of the project: a closer examination of why the turbocharger is failing to support additional airflow and power, how far it could have been pushed before causing damage, and what changes are required to continue increasing manifold air density. The broader lesson is that power development on a diesel is a sequence of density-related limits, and each one has to be measured and understood before moving on to the next.