RunMat
  • Pricing
RunMat
GitHub
GitHub
Sign inTry in browser
DesktopRuntimePhysicsCADAdministrationAgent

Get RunMat updates

New releases, guides and examples.

Product+
  • Runtime
  • GPU
  • Parallel
  • Compile
  • Interop and MEX
  • Simulation
  • MCAD
  • ECAD
  • Enterprise
  • AI integrations
  • Pricing
  • Download
Math+
  • Signal processing
  • Linear algebra
  • Statistics
  • Curve fitting
  • Differential equations
  • Control systems
  • Optimization
  • Finance
  • Data and tables
  • Plotting
Physics+
  • Simulation overview
  • Structural
  • Thermal
  • Electromagnetic
  • Joule heating
  • Thermal stress
  • Acoustics
  • Fluid flow
  • Conjugate heat transfer
  • Fluid–structure
  • What is FEA
Industries+
  • Aerospace
  • Automotive and transportation
  • Defense
  • Electronics and semiconductors
  • Energy
  • Industrial machinery
  • Life sciences
  • Quantitative finance
Learn+
  • Guides
  • Docs
  • Function reference
  • Language reference
  • Migration guide
  • Changelog
  • MATLAB Online
  • MATLAB AI agent
  • For students
Compare+
  • RunMat vs MATLAB
  • RunMat vs Ansys
  • RunMat vs SOLIDWORKS
  • RunMat vs Altium
  • RunMat vs Cadence
  • RunMat vs COMSOL
  • RunMat vs Python
  • RunMat vs Julia
  • RunMat vs Physics AI
  • RunMat vs Octave
  • Benchmarks
  • MATLAB alternatives
  • All comparisons
Company+
  • About
  • Blog
  • Contact
  • Source code↗
  • License
  • Privacy
  • Terms

Product

  • Runtime
  • GPU
  • Parallel
  • Compile
  • Interop and MEX
  • Simulation
  • MCAD
  • ECAD
  • Enterprise
  • AI integrations
  • Pricing
  • Download

Math

  • Signal processing
  • Linear algebra
  • Statistics
  • Curve fitting
  • Differential equations
  • Control systems
  • Optimization
  • Finance
  • Data and tables
  • Plotting

Physics

  • Simulation overview
  • Structural
  • Thermal
  • Electromagnetic
  • Joule heating
  • Thermal stress
  • Acoustics
  • Fluid flow
  • Conjugate heat transfer
  • Fluid–structure
  • What is FEA

Industries

  • Aerospace
  • Automotive and transportation
  • Defense
  • Electronics and semiconductors
  • Energy
  • Industrial machinery
  • Life sciences
  • Quantitative finance

Learn

  • Guides
  • Docs
  • Function reference
  • Language reference
  • Migration guide
  • Changelog
  • MATLAB Online
  • MATLAB AI agent
  • For students

Compare

  • RunMat vs MATLAB
  • RunMat vs Ansys
  • RunMat vs SOLIDWORKS
  • RunMat vs Altium
  • RunMat vs Cadence
  • RunMat vs COMSOL
  • RunMat vs Python
  • RunMat vs Julia
  • RunMat vs Physics AI
  • RunMat vs Octave
  • Benchmarks
  • MATLAB alternatives
  • All comparisons

Company

  • About
  • Blog
  • Contact
  • Source code↗
  • License
  • Privacy
  • Terms
GitHubXLinkedIn

© Dystr, Inc. · RunMat v0.7.1, Nov 17, 2026

RunMat® is a registered trademark of Dystr, Inc. MATLAB® is a registered trademark of The MathWorks, Inc. RunMat is not affiliated with, endorsed by, or sponsored by The MathWorks, Inc.

/
    • Study language
    • Geometry
    • Meshing
    • Materials
    • Models
    • Initial conditions
    • Loads
    • Structural loads
    • Thermal loads
    • Electromagnetic loads
    • Acoustic loads
    • Fluid loads
    • Constraints
    • Structural constraints
    • Thermal constraints
    • Electromagnetic constraints
    • Acoustic constraints
    • Fluid constraints
    • Connectors
    • Contact and interfaces
    • Inputs, sweeps and cases
    • Study contracts
Docs>Physics>Set up

Acoustic loads

Acoustic pressure and normal velocity are conjugate boundary variables. A rigid wall imposes vn=0v_n=0vn​=0; an impedance boundary imposes p=Zvnp=Zv_np=Zvn​; a radiation boundary approximates an outgoing-wave relation. They change the Helmholtz boundary operator in different ways.

The native typed executor supports standalone canonical acoustic harmonic studies; see the retained authored duct evidence. It does not cover every acoustic declaration and is not a general exterior-wave solver.

KindQuantity/fieldsUse
acoustic_pressurePaPressure signal on an acoustic boundary
incident_pressurePaUniform normally incoming plane-wave pressure, not total pressure
normal_velocitym/sDrive a boundary through normal velocity
volume_velocitym³/sTotal outward volume flow over a finite exterior patch
acoustic.rigid_walltargetZero normal velocity
acoustic.radiationtargetOutgoing-wave truncation
acoustic.impedancetarget, impedancep=Zvnp=Zv_np=Zvn​

Every implemented acoustic action and boundary on this page targets a face region. Pressure and normal velocity are scalar normal quantities; they do not accept a separate global direction.

How do I inject incoming pressure without fixing the reflected pressure?

excitation incoming_wave {
    incident_pressure {
        target = geometry.duct.region.inlet
        signal = signal.phased_pressure
    }
}

The signal supplies incoming pressure. Total pressure remains an unknown, allowing a reflected wave to leave through the same face. For peak phasors with exp(+i omega t), the operator uses v_n = Y (p - 2 p_inc), where Y = 1/(rho*c) comes from the unique exterior owning fluid. Reported input power is net incoming power (incoming minus reflected); it is not twice incident power from Robin source-work bookkeeping. The source has the same shared scalar pressure-signal and projection-provenance contract as other excitations, without private amplitude/phase arrays.

This bounded implementation requires coplanar equally oriented exterior faces in one homogeneous medium, uniform distribution, fixed frame and no separate direction. Conflicting sources, natural boundaries, pressure prescriptions and periodic pairing reject. Curved, oblique and spatially varying incidence are not supported by this action. Do not add a separate impedance boundary on its target: the incoming action already includes the matching outgoing operator.

load driver_velocity {
    normal_velocity on geometry.cavity.region.driver = 0.01 m/s
}

load absorbing_end {
    acoustic.impedance {
        target = geometry.duct.region.outlet
        impedance = 415 Pa*s/m
    }
}

How do I apply an acoustic pressure source?

signal source_pressure : pressure over frequency {
    samples = [
        { at = 100 Hz, value = 1 Pa },
        { at = 1000 Hz, value = 2 Pa },
    ]
    interpolation = linear
    extrapolation = reject
}

excitation speaker_pressure {
    acoustic_pressure {
        target = geometry.cavity.region.driver
        signal = signal.source_pressure
    }
}

The signal owns frequency/amplitude data; the excitation owns the boundary target. If phase matters, use the complex signal shape supported by the procedure instead of treating magnitude as a complete complex pressure.

How do I apply acoustic pressure with phase?

Use complex pressure samples. A negative real scalar represents a 180∘180^\circ180∘ phase change; other phases require both components:

signal phased_pressure : pressure over frequency {
    samples = [
        { at = 250 Hz, value = { real = 1 Pa, imaginary = 0 Pa } },
        { at = 500 Hz, value = { real = 0 Pa, imaginary = -1 Pa } },
    ]
    interpolation = linear
    extrapolation = reject
}

excitation phased_driver {
    acoustic_pressure {
        target = geometry.cavity.region.driver
        signal = signal.phased_pressure
    }
}

The second sample has phase −90∘-90^\circ−90∘ under the declared harmonic convention. Preserve that convention when comparing source power, pressure, and particle velocity.

How do I apply a transient pressure pulse?

The signal language can represent pressure over time, but the current typed procedure set has no acoustic transient procedure to consume it. The following declaration is typeable signal/excitation data, not a runnable acoustic analysis today:

signal pressure_pulse : pressure over time {
    samples = [
        { at = 0 ms, value = 0 Pa },
        { at = 1 ms, value = 20 Pa },
        { at = 3 ms, value = 0 Pa },
    ]
    interpolation = linear
    extrapolation = zero
}

excitation pulsed_source {
    acoustic_pressure {
        target = geometry.cavity.region.source_face
        signal = signal.pressure_pulse
    }
}

A future transient procedure must resolve the pulse bandwidth. A short rise time can require a much smaller time step and finer acoustic mesh than its total duration suggests.

How do I drive an acoustic boundary with normal velocity?

load piston_velocity {
    normal_velocity on geometry.cavity.region.piston = 0.01 m/s
}

Normal velocity models piston/diaphragm motion without prescribing pressure. Its sign follows the resolved outward normal. For coupled structural motion, use FSI/acoustic-structure transfer rather than manually duplicating a velocity trace.

How do I prescribe total volume flow rather than speed?

signal total_flow : volume_flow_rate over constant { value = -0.01 m^3/s }

excitation volume_driver {
    volume_velocity {
        target = geometry.cavity.region.piston
        signal = signal.total_flow
    }
}

volume_velocity uses the shared scalar volume_flow_rate signal. The total is distributed once over the complete selected exterior patch as v_n=Q/A, not once per face. Its sign is outward-positive; a negative value injects flow into the domain. m/s and kg/s signals are different quantities and reject. There is no separate direction, follower motion or moment projection. Complex frequency samples retain both components through the ordinary shared signal contract.

This bounded CPU implementation is under qualification; see the volume-flow evidence and remaining gates. It is a finite boundary source, not an interior monopole or an arbitrary spatial distribution. Do not overlap another acoustic source, pressure prescription, rigid/impedance boundary or paired periodic relation on its patch.

How do I apply a frequency-dependent piston velocity?

Use a velocity signal over frequency and an explicit normal-velocity excitation:

signal piston_spectrum : velocity over frequency {
    samples = [
        { at = 50 Hz, value = 0.2 mm/s },
        { at = 500 Hz, value = 1.0 mm/s },
        { at = 2 kHz, value = 0.3 mm/s },
    ]
    interpolation = linear
    extrapolation = reject
}

excitation piston_drive {
    normal_velocity {
        target = geometry.cavity.region.piston
        signal = signal.piston_spectrum
    }
}

The scalar is resolved along the acoustic boundary normal. Reverse geometry orientation or the signal sign deliberately; do not add an arbitrary global direction to a normal-velocity condition.

How do I model a rigid acoustic wall?

load enclosure_wall {
    acoustic.rigid_wall { target = geometry.cavity.region.walls }
}

A rigid wall sets normal velocity to zero. It reflects acoustic energy; it is not absorbing. Use it only where wall compliance and loss are negligible for the frequency range of interest.

How do I model an open or radiating acoustic boundary?

load open_end {
    acoustic.radiation { target = geometry.duct.region.opening }
}

Radiation approximates outgoing waves at a truncated domain. Place the boundary where its admitted approximation is valid and test domain-size sensitivity. It is not interchangeable with a measured termination impedance or a perfectly matched layer.

How do I apply an acoustic impedance boundary?

load liner {
    acoustic.impedance {
        target = geometry.duct.region.lined_wall
        impedance = 500 Pa*s/m
    }
}

Specific acoustic impedance relates pressure to normal velocity, p=Zvnp=Zv_np=Zvn​, with vnv_nvn​ along the outward normal. A quantity is a constant real (resistive) impedance; a complex or frequency-dependent impedance comes from a table (below). Check that the value is specific impedance rather than total mechanical impedance.

How do I model a resistive acoustic liner?

Use acoustic.impedance with the measured or derived specific impedance in Pa*s/m. For a purely resistive liner, a positive real value dissipates energy under the formulation's normal/sign convention. Apply it only to the lined faces:

load resistive_liner {
    acoustic.impedance {
        target = geometry.duct.region.liner_faces
        impedance = 650 Pa*s/m
    }
}

Do not multiply specific impedance by area before entering it. Mechanical impedance in N*s/m and surface-specific acoustic impedance are not interchangeable.

How do I model a frequency-dependent complex impedance?

Declare a top-level table of impedance over frequency and reference it from the boundary. A measured liner is a list of resistance and reactance samples; harmonic solves interpolate them linearly in frequency and hold the end values beyond the first and last sample:

table liner_data : impedance over frequency {
    samples = [
        { at = 125 Hz, resistance = 420 Pa*s/m, reactance = -900 Pa*s/m },
        { at = 250 Hz, resistance = 430 Pa*s/m, reactance = -380 Pa*s/m },
        { at = 500 Hz, resistance = 440 Pa*s/m, reactance = -120 Pa*s/m },
    ]
}

model lined_duct {
    load liner {
        acoustic.impedance {
            target = geometry.duct.region.liner_faces
            impedance = table.liner_data
        }
    }
}

A rigidly backed porous layer (mineral wool, foam, fibrous ceiling tiles) is described by its flow resistivity and thickness with the Delany–Bazley model, which gives the layer's characteristic impedance and wavenumber in the adjacent air; the surface impedance is Z=−iZccot⁡(kcd)Z=-iZ_c\cot(k_cd)Z=−iZc​cot(kc​d):

table ceiling_tile : impedance over frequency {
    model = delany_bazley
    flow_resistivity = 20000 Pa*s/m^2
    thickness = 50 mm
}

Reactance follows the e+iωte^{+i\omega t}e+iωt convention of the harmonic solver: a stiffness-like (spring) surface has negative reactance and a mass-like surface positive reactance. Resistance must be nonnegative. Impedance tables apply to harmonic studies; modal and transient studies take rigid walls, pressure release and constant impedance.

How do I model a perfectly matched layer (PML)?

acoustic.radiation is a boundary approximation, not a PML. A PML requires a transformed exterior volume, thickness and coordinate-stretch law. The current closed load set has no first-class PML object. Extend the acoustic formulation with a typed PML domain before claiming PML behavior; do not rename a radiation boundary.

How do I model a loudspeaker or vibrating diaphragm?

Use normal_velocity when diaphragm velocity is known, or an acoustic_pressure excitation when the boundary pressure is known. If diaphragm motion must be solved from structural dynamics, use an admitted acoustic-structure/FSI transfer instead of imposing the same guessed motion on both physics. The sign of normal velocity follows the acoustic domain's outward normal.

How do I model structure-borne sound from a vibrating panel?

Use a coupled acoustic-structure interface so structural normal velocity is transferred to the acoustic boundary and acoustic pressure returns traction to the structure when two-way coupling is admitted. A copied normal_velocity trace is a one-way imposed approximation; if used, document the source run and topology mapping as evidence rather than presenting it as coupled response.

The current interface catalog exposes a fluid-structure interface, not a dedicated acoustic-structure interface declaration. Do not present that interface as acoustic coupling without a procedure that explicitly binds structural and acoustic states.

How do I model a monopole or point acoustic source?

The current public load set is boundary-based and does not expose a point-monopole or volume source strength. Smearing a point source over an arbitrary tiny face changes its units and mesh dependence. A first-class source needs a typed volume-velocity/source-strength quantity, location or region, regularization, and power-balance evidence.

How do I model an incident plane wave?

An incident-field/scattered-field decomposition needs propagation direction, amplitude, phase, medium properties, and a compatible truncation boundary. It is not equivalent to uniform pressure on every exterior face. The current closed declarations do not expose an incident-plane-wave object; use an admitted formulation with that typed contract or author the physically equivalent boundary data only when it is known on the entire truncation surface.

How do I model diffuse-field or random acoustic excitation?

A diffuse field is a spatially correlated stochastic field, not one scalar pressure history. It requires cross-spectral density/coherence and directional sampling. Do not attach the same random signal independently to every face. Use a stochastic acoustic procedure only when it preserves the correlation model and reports the convention used.

How do I model an anechoic termination or open end?

Use acoustic.radiation only when its outgoing-wave approximation is appropriate at the truncation surface. Use acoustic.impedance for a measured or designed termination relation. A rigid wall reflects energy and is the opposite of an anechoic termination. Check reflection sensitivity by moving the truncation boundary or comparing against a known duct mode.

How do I apply a sound-pressure level in decibels?

The typed load quantity is pressure in Pa, not dB. Convert an RMS sound-pressure level using prms=pref10Lp/20p_{rms}=p_{ref}10^{L_p/20}prms​=pref​10Lp​/20 with the correct reference pressure for the medium, then state whether the harmonic solver expects RMS, peak, or complex amplitude. Do not enter the dB number as pascals.

For air, prefp_{ref}pref​ is commonly 20 μPa20\,\mu\text{Pa}20μPa; other media and standards can use a different reference. A level alone still does not state phase or spatial coherence. Record amplitude convention and reference pressure in the study inputs/evidence.

How do I combine several speakers or acoustic sources?

Give each independently controlled source its own signal and excitation, include them in an explicit load case, and preserve their relative phases. Adding scalar magnitudes loses interference. For uncorrelated random sources, combine response spectra according to the stochastic procedure rather than assigning arbitrary deterministic phase.

Why is the acoustic source producing the wrong sign or phase?

Inspect the resolved face normal, source sign, harmonic e+iωte^{+i\omega t}e+iωt or e−iωte^{-i\omega t}e−iωt convention, and primary-to-secondary orientation of any periodic boundary. A normal-velocity sign error can look like a 180-degree phase error. Verify against a simple duct or cavity solution before interpreting a complex assembly.

What should I inspect after applying acoustic loads?

Inspect resolved target area, normal orientation, input pressure/velocity convention, achieved boundary resultant or power, reflected flux at truncations, and global acoustic energy balance. For frequency sweeps, inspect interpolation coverage and confirm no solver sample relied on rejected or unintended extrapolation.

How do I prescribe acoustic power?

The current closed load set has pressure and normal-velocity drives, not a direct total acoustic- power declaration. Converting watts to pressure requires radiation area, impedance, mode shape, and amplitude convention. Author one of the supported boundary variables only when those assumptions are known; otherwise a first-class resultant acoustic-power source is required.

Use an excitation with a pressure or velocity signal for transient or harmonic drive. A rigid wall, radiation boundary, and impedance boundary describe different physical terminations; none should be selected from geometry names automatically.

How do I convert sound-pressure level to an acoustic pressure input?

For a stated reference pressure prefp_{ref}pref​ and RMS convention, prms=pref10Lp/20p_{rms}=p_{ref}10^{L_p/20}prms​=pref​10Lp​/20. Convert to peak only if the harmonic procedure expects peak amplitude. Then apply the pressure with its phase as an acoustic_pressure signal. State whether the level is free-field, diffuse-field, incident, or total pressure; the same decibel value does not define the boundary wave decomposition.

How do I model a duct inlet and anechoic outlet?

Drive the inlet with normal_velocity or acoustic_pressure and terminate the outlet with acoustic.radiation or a characterized acoustic.impedance. Keep wall regions under acoustic.rigid_wall unless wall compliance is modeled through an FSI interface. The outlet must be far enough from strong evanescent content for the selected termination law to be valid.

load duct_inlet {
    normal_velocity on geometry.duct.region.inlet = 0.02 m/s
}

load open_end {
    acoustic.radiation {
        target = geometry.duct.region.outlet
    }
}

How do I model a cavity resonance test?

Use rigid or impedance wall boundaries matching the test article, drive a small known pressure or normal velocity on the physical source patch, and sweep frequency with sufficient resolution around expected modes. Request complex pressure at microphone or probe locations and wall/source power. A uniform pressure constraint on the entire cavity suppresses rather than excites many cavity modes.

How do I verify acoustic energy balance?

Compare time-averaged input power at driven boundaries with radiated, absorbed, and dissipated power. Inspect reflected flux at truncations and phase at probes, not only pressure magnitude. Mesh and frequency refinement must resolve wavelength and boundary layers required by the formulation; a stable-looking SPL plot with nonclosing power balance is not sufficient evidence.

How do I choose between pressure and normal-velocity excitation?

Use acoustic_pressure when boundary pressure is the controlled source. Use normal_velocity when the normal motion of a piston, diaphragm, or wall is known. Pressure is effort-like and normal velocity is flow-like; prescribing both on the same boundary generally overconstrains the acoustic port.

For a source face Γ\GammaΓ, time-averaged complex power is Pˉ=12Re⁡∫Γp vn∗ dΓ\bar P=\tfrac12\operatorname{Re}\int_\Gamma p\,v_n^*\,d\GammaPˉ=21​Re∫Γ​pvn∗​dΓ. Request both conjugate quantities from the solution even when only one is prescribed, then check sign and delivered power.

How do I drive a speaker diaphragm from measured motion?

Convert the measured normal surface velocity into a typed velocity signal, retaining whether the measurement is peak, RMS, or complex amplitude. Apply it only to the radiating diaphragm region.

signal diaphragm_velocity : velocity over frequency {
    samples = [
        { at = 100 Hz, value = { real = 0.01 m/s, imaginary = 0 m/s } },
        { at = 1000 Hz, value = { real = 0.004 m/s, imaginary = -0.003 m/s } },
    ]
    interpolation = linear
    extrapolation = reject
}

excitation speaker_motion {
    normal_velocity {
        target = geometry.enclosure.region.diaphragm
        signal = signal.diaphragm_velocity
    }
}

Confirm face-normal orientation and compare integrated volume velocity Q=∫Γvn dΓQ=\int_\Gamma v_n\,d\GammaQ=∫Γ​vn​dΓ with the measurement. A single uniform value is an approximation when the diaphragm bends. The current .fea excitation grammar does not accept a spatial complex velocity field, so measured nonuniform motion needs a future topology-field contract that preserves coordinates and phase.

How do I model a muffler, silencer, or lined duct?

Drive the inlet with pressure or normal velocity, apply acoustic.impedance to the lined faces, and terminate the outlet with acoustic.radiation or a validated impedance. Keep rigid casing walls as acoustic.rigid_wall. Compute incident and transmitted power or a two-port quantity; pressure level at one point is not insertion loss.

Check wavelength resolution, liner impedance convention, reflected power at the outlet, and energy absorbed by the liner. If mean flow materially changes propagation, use a formulation that owns that coupling rather than modifying sound speed by hand.

How do I model two coherent acoustic sources?

Create one excitation and complex signal per independently controlled source. Their phases determine interference. A 180-degree phase difference is represented by the complex signal, not by reversing the target face normal. Include both excitations in the same case and inspect complex pressure and power before converting to SPL.

For incoherent or partially coherent sources, deterministic phasor addition is not valid. The study needs cross-spectral density or an admitted random-acoustic contract; do not approximate incoherence by choosing arbitrary phases.

How do I verify a one-dimensional duct source and termination?

Use a straight, constant-area duct with a source at one end and the chosen radiation or impedance condition at the other. Request complex pressure and normal velocity on at least two sections. Check the time-averaged power

Pˉ=12Re⁡∫Γpvn∗ dΓ\bar P=\tfrac12\operatorname{Re}\int_\Gamma p v_n^*\,d\GammaPˉ=21​Re∫Γ​pvn∗​dΓ

and compare the measured reflection coefficient with the analytic termination value. Repeat with a finer mesh and a longer duct or moved truncation plane. Pressure magnitude alone cannot distinguish a correct traveling wave from a standing wave produced by the wrong normal, impedance, or phasor sign convention.

Further reading: acoustic constraints, acoustic studies, and fluid–structure interaction.

On this page
  • How do I inject incoming pressure without fixing the reflected pressure?
  • How do I apply an acoustic pressure source?
  • How do I apply acoustic pressure with phase?
  • How do I apply a transient pressure pulse?
  • How do I drive an acoustic boundary with normal velocity?
  • How do I prescribe total volume flow rather than speed?
  • How do I apply a frequency-dependent piston velocity?
  • How do I model a rigid acoustic wall?
  • How do I model an open or radiating acoustic boundary?
  • How do I apply an acoustic impedance boundary?
  • How do I model a resistive acoustic liner?
  • How do I model a frequency-dependent complex impedance?
  • How do I model a perfectly matched layer (PML)?
  • How do I model a loudspeaker or vibrating diaphragm?
  • How do I model structure-borne sound from a vibrating panel?
  • How do I model a monopole or point acoustic source?
  • How do I model an incident plane wave?
  • How do I model diffuse-field or random acoustic excitation?
  • How do I model an anechoic termination or open end?
  • How do I apply a sound-pressure level in decibels?
  • How do I combine several speakers or acoustic sources?
  • Why is the acoustic source producing the wrong sign or phase?
  • What should I inspect after applying acoustic loads?
  • How do I prescribe acoustic power?
  • How do I convert sound-pressure level to an acoustic pressure input?
  • How do I model a duct inlet and anechoic outlet?
  • How do I model a cavity resonance test?
  • How do I verify acoustic energy balance?
  • How do I choose between pressure and normal-velocity excitation?
  • How do I drive a speaker diaphragm from measured motion?
  • How do I model a muffler, silencer, or lined duct?
  • How do I model two coherent acoustic sources?
  • How do I verify a one-dimensional duct source and termination?