Diffusers

The diffuser is the first physical element air encounters when it enters the combustor. Its job is simple to state and surprisingly hard to do well: slow the air from compressor exit conditions (~150 m/s) down to something the combustor can work with (~20–25 m/s), while recovering as much of the kinetic energy as possible as static pressure.

Why does this matter? Because pressure is the resource a gas turbine runs on. Every pascal of stagnation pressure lost in the diffuser is permanently unavailable to the turbine — it reduces thrust and increases fuel consumption. At the same time, if you slow the air poorly (separating the boundary layer, creating recirculation within the diffuser itself), you feed the combustion zone non-uniform, turbulent flow that complicates flame stability.

The physics of deceleration

Bernoulli's equation describes the ideal case: increase the cross-sectional area of the flow path, and the velocity drops while pressure rises. The stagnation pressure (total energy) is conserved.

In the real world, it is not. Viscosity creates a boundary layer on the diffuser walls. If the pressure gradient is too steep — if the flow tries to decelerate too quickly — the boundary layer cannot sustain the adverse pressure gradient and separates from the wall. The separated region becomes a low-energy recirculation zone, and the effective flow area collapses. The result is high total pressure loss, unsteady flow, and potentially severe flow non-uniformity entering the combustor.

Performance metrics

Two coefficients capture diffuser quality:

Static pressure recovery coefficient $C_p$:

$C_p = \frac{p_2 - p_1}{q_1} = \frac{p_2 - p_1}{\frac{1}{2}\rho_1 V_1^2}$

This measures what fraction of the inlet dynamic pressure was successfully converted to static pressure. An ideal (isentropic) diffuser achieves $C_p = 1 - (A_1/A_2)^2$. Real diffusers fall short.

Total pressure loss coefficient $\lambda$:

$\lambda = \frac{p_{t1} - p_{t2}}{q_1}$

This measures what fraction of the inlet dynamic pressure was lost to viscous dissipation and turbulence. Good diffuser design simultaneously maximizes $C_p$ and minimizes $\lambda$.

Faired diffusers

A faired (or aerodynamic) diffuser is a smoothly contoured expanding duct. The walls follow the flow closely, guiding it to decelerate without separation.

The classic result from diffuser theory: for a plane two-dimensional diffuser, the maximum total divergence angle before separation occurs is around 7°. For axisymmetric (conical) diffusers, around 10°. Steeper than this and the boundary layer detaches.

The implication is physically intuitive: if you need to expand the flow by a factor of 5 (typical for a combustor), and you can only use 7° half-angles, the diffuser must be quite long. For an aircraft engine where every centimeter of length adds weight and drag, "quite long" is a real problem.

When faired diffusers are used: In industrial and marine gas turbines where weight is less critical, and where the high pressure-recovery efficiency justifies the length. Also as short pre-diffusers upstream of a dump stage.

Dump diffusers

The dump diffuser abandons the goal of perfectly guided flow in exchange for radical shortness. It consists of a short pre-diffuser that reduces the Mach number from ~0.3 to around 0.1–0.15, followed by an abrupt area expansion — the "dump" — into the annular combustor space.

What happens at the dump? The flow separates and forms a large recirculation zone in the corners of the casing. Instead of fighting this — as a faired diffuser tries to do — the dump diffuser design deliberately accepts this corner recirculation as a stable, predictable feature. The recirculating fluid acts as a cushion, and the core jet decelerates by spreading into the larger space.

Why dump diffusers won:

1. Length: A dump diffuser can accomplish the required deceleration in a fraction of the axial length of a faired design.

2. Robustness to swirl: Compressor exit air carries residual swirl (it has been through rotating blades). A faired diffuser is highly sensitive to inlet flow angle — swirl can cause asymmetric separation and performance degradation. A dump diffuser is relatively insensitive; the sudden expansion overwhelms any inlet non-uniformity.

3. Stability: The corner vortices are topologically stable. They do not move or flap the way a separated boundary layer on a faired wall does.

The cost: Slightly higher total pressure loss than a theoretically ideal faired diffuser. In practice, the loss difference is small enough that the weight and robustness advantages decisively favor the dump design for flight applications.

The pre-diffuser as first stage

Modern designs often use a two-stage approach: a short faired pre-diffuser followed by a dump. The pre-diffuser reduces velocity moderately without risking separation, so that the Mach number at the dump is lower and the sudden expansion is less severe. This captures most of the efficiency of the faired design while retaining the short length and robustness of the dump.

Diffuser design and the combustor interface

The exit conditions of the diffuser directly set the conditions at the combustor entry — velocity, total pressure, and flow uniformity. A diffuser that creates a clean, uniform, low-velocity flow field makes the combustor designer's job tractable. A diffuser that introduces large-scale unsteadiness or flow asymmetry forces the combustor to absorb that non-uniformity, which typically shows up as poor pattern factor or stability margin.

In this sense the diffuser and the combustor are not independent subsystems. Their interface must be co-designed, and the combined performance is what matters — not the performance of either element in isolation.

In Lesson 4, we move into the combustor itself and look at the aerodynamics that actually hold the flame: swirlers, recirculation zones, and the carefully choreographed system of air jets that sculpt the flow field inside the liner.