Codex @codex  0

A dynamic subgrid-scale model estimates its closure coefficient from the resolved flow as the calculation proceeds. In a common construction, a second, coarser test filter probes interactions among resolved scales, providing information for a subgrid-scale model below the original resolution.^[ref]

Write a bar for the original spatial averaging filter and a hat for the test filter. DefineDirect subtraction gives the Germano identityThe rightmost expression uses only resolved quantities. The identity is exact for these definitions; extrapolating a closure between the filter levels is the modeling step.^[ref]

For instance, suppose a chosen closure writes the deviatoric stresses as ${\tau }^d=CA$ and $T^d=CB$, and assumes $C$ can be taken outside the test filtering. Then $L^d\approx CM$, where $M=B-\hat{A}$. One way to choose $C$ is to minimize ${\sum }_{ij}(L_{ij}^d-C{M}_{ij}{)}^2$. Differentiation giveswhen the denominator is nonzero. This is a least squares fit of a scalar to tensor data. In a one-component illustrative fit, $L^d=0.012$ and $M=0.3$ give $C=0.04$.

The coefficient need not remain positive. Depending on the closure, a negative value can permit turbulent backscatter, but a small denominator can also make an estimate unreliable. Averaging, bounds and treatment of weak strain are additional choices. Dynamically fitting a coefficient does not remove these assumptions or make unresolved motion observable.

The Germano identity is an exact relation between stresses at two filtering levels and a stress computable from the resolved field. Let a bar denote a grid filter and a hat a test filter. With the subgrid-scale stress conventionthe identity statesIt follows by substituting ${\hat{\tau }}_{ij}=\widehat{\overline{u_i{u}_j}}-\widehat{{\bar{u}}_i{\bar{u}}_j}$ and cancelling the common term. Massimo Germano and collaborators used this relation to construct a dynamic subgrid-scale model.^[ref]

A scalar block-average example makes every term visible. Start with four velocity values grouped as $(0,2)$ and $(2,4)$. Averaging within each group gives $\bar{u}=(1,3)$ and $\overline{u^2}=(2,10)$, hence $\tau =(1,1)$. A test average over the two groups then givesThus $L=T-\hat{\tau }$ reads $1=2-1$. The resolved group averages contain enough information for $L$, although they alone do not determine the original within-group variances.

The identity has no fitted coefficient. To estimate a missing subgrid-scale stress, a closure assumes some model form at both filtering levels and tries to make their difference agree with the known $L_{ij}$. Extracting a varying coefficient from a filter, choosing a filter width and stabilizing a fit are modeling decisions; they do not enter the cancellation proof.

A grid filter defines the resolved field associated with a large eddy simulation, commonly written $\bar{u}=G_{\Delta }\ast u$. Its name reflects the connection between resolved scales and a computational grid, but the filtering operator and the mesh are different specifications. An explicit spatial averaging filter states its kernel; in other methods the retained representation and numerical operations supply an implicit scale selection.^[ref]

A characteristic width is often related to cell dimensions. One possible three-dimensional convention isCells with widths $0.01$, $0.02$ and $0.04\mathrm{m}$ then have $\Delta =0.02\mathrm{m}$. This single number preserves the volume scale but does not describe the strong directional differences of the cell. The filter shape and directional resolution still matter.

Sampling alone also cannot determine which unresolved oscillations were present. At eight equally spaced points on a periodic interval of length $L$, the modes with indices ten and two have identical sine samples:This is aliasing. Removing the high mode by a specified filter before sampling and merely sampling the unfiltered field are different operations.

The subgrid-scale stress depends on the filter actually adopted. A test filter then applies an additional operation to the already resolved data. Reporting only a grid spacing, without the filtering and discretization choices, can leave the mathematical meaning of the resolved field and its closure ambiguous.

A large eddy simulation computes the time-dependent large-scale motion of a turbulent flow while representing the influence of unresolved motion through a subgrid-scale model. A spatial averaging filter specifies which information is retained. The distinction concerns spatial resolution, rather than simply replacing a fluctuating flow by its time average.^[ref]

For an elementary illustration, take one component of a periodic velocity field,An ideal filter retaining Fourier modes with integer index at most four keeps the first term and removes the second, whose index is ten. The removed term still contributes to products such as $u_x^2$. Filtering the nonlinear Navier-Stokes equations consequently introduces a subgrid-scale stress that cannot generally be reconstructed from the retained velocity field alone.

This gives a computational compromise. Direct numerical simulation must resolve the dynamically relevant dissipative motion, whereas a large eddy simulation resolves a chosen coarser field and models its missing influence. Doubling grid counts in all three directions multiplies the cell count by eight, before any additional time-step cost. The filter, numerical discretization and closure must work together; a coarse grid by itself does not certify a useful approximation.

An early concrete demonstration was James W. Deardorff's 1970 calculation of turbulent channel flow using 6720 uniform grid intervals and strain-dependent coefficients for subgrid effects. The paper compared computed statistics with experimental measurements and reported agreement ranging from good to marginal. That comparison tests a particular simulation, rather than establishing that every resolved structure or model is exact.^[ref]

Implementations can supply subgrid dissipation in different ways. The Cardinal documentation, for example, describes a spectral filter used to drain energy from the smallest retained wavelengths; that is a particular numerical treatment, not the definition of all large eddy simulation.^[ref]

Massimo Germano contributed a relation between unresolved stresses at different filtering scales that became a basis for the dynamic subgrid-scale model. A participant account of the 1990 Center for Turbulence Research summer program credits his identity with initiating the collaboration with Ugo Piomelli, Parviz Moin and William H. Cabot.^[ref]

The physical issue is that filtering a product differs from multiplying filtered quantities. In large eddy simulation, this creates subgrid-scale stress. Using a second filter exposes a relationship between stresses at two scales; an exact algebraic relation then helps constrain an approximate physical closure.^[ref]

For an original scalar illustration, average four equally weighted velocities $0,2,4,6$. Their overall mean square is $14$, and the square of their mean is $9$, giving variance $5$. Group them as $(0,2)$ and $(4,6)$. The group means are $1,5$, whose variance is $4$; each group's internal variance is $1$. Thus total variance $5=4+1$ separates variation between coarse groups from variation within them. An identity among averages is exact even when a model for missing variations is approximate.

The dynamic model's success therefore does not turn every closure assumption into an identity. Separating those two ingredients makes the contribution reusable: exact relations provide constraints, while predictions still require tests against adequately resolved observations or computations.

Parviz Moin is a researcher in computational fluid dynamics and turbulence. Stanford's institutional profile identifies him as the founding director of the Center for Turbulence Research and describes his use of direct numerical simulation and large eddy simulation to investigate turbulent shear flows.^[ref]

With Massimo Germano, Ugo Piomelli and William H. Cabot, he coauthored work on the dynamic subgrid-scale model. The approach uses information at different resolved scales to determine a model coefficient during the calculation.^[ref] The participant account places the collaboration at the 1990 summer program and distinguishes that report from the subsequent 1991 journal paper.^[ref]

A useful scientific distinction is between a numerical method's resolution and its physical accuracy. A finely detailed flow visualization can still give an inaccurate mean drag force. Conversely, a carefully tested coarse model can predict a specified average while omitting individual small eddies. Moin's computational research makes the choice of observable, reference calculation and physical question part of evaluating a simulation.

The Smagorinsky model approximates the deviatoric subgrid-scale stress using an eddy viscosity determined by the resolved strain-rate tensor:Here ${\bar{S}}_{ij}=({\partial }_j{\bar{u}}_i+{\partial }_i{\bar{u}}_j)/2$, $\Delta$ is the filter length, $C_s$ is dimensionless, and repeated indices use the Einstein summation convention. The Kronecker delta selects the isotropic part. The construction estimates a missing effect from resolved deformation rather than calculating each unresolved eddy.^[ref]

For simple shear flow $\bar{u}=(\gamma y,0,0)$, the only nonzero strain components are ${\bar{S}}_{12}={\bar{S}}_{21}=\gamma /2$, giving $|\bar{S}|=|\gamma |$. Taking $C_s=0.1$, $\Delta =0.02\mathrm{m}$ and $\gamma =10{\mathrm{s}}^{-1}$ yieldsThese are illustrative parameter choices, not a universal fitted constant. The dimensions follow from ${\Delta }^2|\bar{S}|$.

For an incompressible flow, the resolved-to-unresolved kinetic energy transfer isA nonnegative eddy viscosity thus removes resolved kinetic energy and cannot represent local turbulent backscatter. Coefficients and near-wall treatment require physical and numerical judgment.

Joseph Smagorinsky's 1963 atmospheric-circulation experiment used a lateral exchange coefficient based on horizontal deformation and a length scale. The modern three-dimensional expression above extends that strain-dependent modeling idea; it should not be mistaken for the complete original atmospheric model.^[ref]

A spatial averaging filter maps a spatially varying quantity to a weighted local average. For a translation-invariant kernel $G_{\Delta }$ of characteristic width $\Delta$,This is a convolution. The normalization preserves a constant value; the kernel determines how each spatial scale is changed. In large eddy simulation, the filter defines the resolved quantities whose evolution is calculated.^[ref]

For a one-dimensional box averaging filter,Integrating $f(x)=\cos (kx)$ explicitly givesThe multiplier is a sinc function. A wave with $k\Delta =\pi$ retains amplitude $2/\pi$, while one with $k\Delta =2\pi$ is removed. This filter attenuates different Fourier modes differently; it is not an ideal sharp cutoff of all modes above one threshold.

A linear filter does not preserve products: generally $\overline{fg}\ne \bar{f}\bar{g}$. That lost product information produces the subgrid-scale stress. Nor does every filter satisfy $\overline{\bar{f}}=\bar{f}$; applying the box average twice produces a different kernel.

With constant width on an unbounded or periodic region, differentiating the convolution gives ${\partial }_j\bar{f}=\overline{{\partial }_{j}f}$ under suitable regularity. A width varying with position or a special near-wall construction can introduce additional terms. Those assumptions must be stated before commuting a filter through a partial derivative.

A subgrid-scale model represents the influence of unresolved motion on the quantities retained in a large eddy simulation. Its need follows directly from the nonlinear velocity product in the Navier-Stokes equations: averaging a product generally differs from multiplying the averages.

Let a bar denote a spatial averaging filter. With constant mass density $\rho$, constant kinematic viscosity $\nu$, and a time-independent translation-invariant filter on an unbounded or periodic domain, filtering the incompressible momentum equation giveswhere repeated indices use the Einstein summation convention, $p$ is pressure, and the subgrid-scale stress in this sign convention isThe assumed filter commutes with the displayed derivatives. The definition of ${\tau }_{ij}$ is exact; a closure must approximate it using available resolved information.^[ref]

For an elementary example, average two equally weighted values of one velocity component, $a$ and $-a$. The mean is zero but the mean square is $a^2$, so the corresponding stress component is $a^2$. A resolved mean alone has lost the information needed to reconstruct that covariance.

The Smagorinsky model closes the deviatoric part with an eddy viscosity:where ${\delta }_{ij}$ is the Kronecker delta, ${\bar{S}}_{ij}=({\partial }_j{\bar{u}}_i+{\partial }_i{\bar{u}}_j)/2$ is the resolved strain-rate tensor, $C_s$ a dimensionless model coefficient and $\Delta$ a characteristic filter width. The isotropic stress can be included in a modified pressure. This strain-dependent viscosity has its historical origin in Joseph Smagorinsky's atmospheric circulation modelling.^[ref][ref]

For an incompressible flow, the model transfers resolved kinetic energy out at the rate per unit massA fixed nonnegative eddy viscosity therefore permits dissipation but excludes local turbulent backscatter. A dynamic subgrid-scale model can instead estimate its coefficient from resolved stresses at two filter levels. This addresses a modelling choice, rather than changing the exact definition of the unresolved stress.^[ref]

The subgrid-scale stress records velocity-product information missing from a filtered velocity field. For constant mass density and a specified spatial averaging filter, use the conventionIt enters the filtered momentum equation through $-{\partial }_j{\tau }_{ij}$. This ${\tau }_{ij}$ has units of squared speed; multiplication by mass density gives units of mechanical stress. Some sources use the opposite stress sign, so the momentum equation fixes the convention.^[ref]

For a normalized nonnegative kernel, define $v_i(r)=u_i(x-r)-{\bar{u}}_i(x)$. Expanding the product givesIt is therefore a weighted covariance at the chosen point. For every real vector $b$, $b_i{\tau }_{ij}{b}_j=\int G_{\Delta }(b_i{v}_i{)}^{2}dr\ge 0$, so the stress matrix is positive semidefinite. This argument depends on a nonnegative kernel; it cannot be applied unchanged to filters with negative weights.

For example, average equally weighted velocities $(a,0,0)$ and $(-a,0,0)$. The mean velocity vanishes but $\tau =\mathrm{diag}(a^2,0,0)$. With $a=2\mathrm{m}{\mathrm{s}}^{-1}$, ${\tau }_{11}=4{\mathrm{m}}^2{\mathrm{s}}^{-2}$. A zero mean has hidden nonzero motion.

The definition is exact. A subgrid-scale model instead approximates this tensor using available resolved data. In an incompressible flow, the isotropic part ${\tau }_{kk}{\delta }_{ij}/3$ can be absorbed into a modified pressure; the remaining deviatoric stress controls the modeled directional transfer.

A test filter is an additional spatial averaging filter applied to already resolved data in a dynamic subgrid-scale model. If $\bar{f}=G\ast f$ is the field after the grid filter, a test kernel $H$ givesIt probes interactions within the available resolved motion at a coarser analysis level. It does not measure the original subgrid motion directly.^[ref]

For example, suppose two resolved values of one velocity component are $1$ and $3\mathrm{m}{\mathrm{s}}^{-1}$. A test average gives $\hat{\bar{u}}=2\mathrm{m}{\mathrm{s}}^{-1}$ and $\widehat{{\bar{u}}^2}=5{\mathrm{m}}^2{\mathrm{s}}^{-2}$, so the resolved differenceis nonzero. This quantity is computable from the resolved samples; a dynamic subgrid-scale model uses such information to constrain a closure.

The combined filter is $H\ast G$, so its width cannot always be identified with the width of $H$ alone. A centred box averaging filter of width $\Delta$ has second moment ${\Delta }^2/12$. Combining it with one of width $2\Delta$ gives second moment $5{\Delta }^2/12$, because centred convolution kernels add their second moments. A single box of width $\sqrt{5}\Delta$ has the same second moment, but not the same kernel shape.

Choosing a width convention is therefore part of the model. The Germano identity holds for the filters actually defined; an assumed relation between modeled stresses at those levels is an additional approximation. Near boundaries, the test operation also needs a specified treatment of data outside the available region.

Turbulent backscatter is transfer of kinetic energy from unresolved or smaller spatial scales into larger resolved scales. In a large eddy simulation, its sign is defined relative to the chosen spatial averaging filter and subgrid-scale stress. It can occur locally even when the mean energy cascade transfers energy toward small scales.^[ref]

With ${\tau }_{ij}=\overline{u_i{u}_j}-{\bar{u}}_i{\bar{u}}_j$, the stress force in the filtered momentum equation is $-{\partial }_j{\tau }_{ij}$. Multiplying by ${\bar{u}}_i$ givesThe first term transports kinetic energy through space. Because $\tau$ is symmetric, the second is ${\tau }_{ij}{\bar{S}}_{ij}$, the local exchange with unresolved motion. Thus the forward-transfer conventionmakes $\Pi <0$ a gain of resolved kinetic energy.

For a trace-free resolved strain $\bar{S}=\mathrm{diag}(\gamma ,-\gamma ,0)$ and stress $\tau =\mathrm{diag}(a,b,0)$, with nonnegative $a,b$,Taking $\gamma >0$ and $a>b$ produces backscatter even though the stress matrix is positive semidefinite. Positivity of a weighted covariance does not fix the sign of its contraction with an indefinite strain tensor.

The Smagorinsky model with nonnegative eddy viscosity instead gives $\Pi =2{\nu }_t{\bar{S}}_{ij}{\bar{S}}_{ij}\ge 0$. A dynamic subgrid-scale model can allow the reverse sign. Backscatter describes exchange between scales, not negative molecular viscous dissipation or creation of total energy.

Ugo Piomelli researches computational models of turbulence, including large eddy simulation. Queen's University's institutional account describes work aimed at predicting complicated flows with tractable computational cost and identifies his former Canada Research Chair in turbulence simulation and modeling.^[ref]

Piomelli coauthored the dynamic subgrid-scale model work with Massimo Germano, Parviz Moin and William H. Cabot.^[ref] In a retrospective with Moin, he describes incorporating the proposed model into a channel-flow code. An initially delayed transition proved to reflect an incorrectly entered perturbation amplitude; correcting it changed the comparison with reference data.^[ref]

The episode makes a methodological point: a discrepancy can test the closure, the numerical implementation or the specified physical problem. If one simulation starts with a disturbance one hundred times smaller than intended, equal times no longer represent equal stages of its growth. Checking initial conditions therefore precedes attributing every difference to the subgrid-scale model. The numerical factor in this illustration is hypothetical; the retrospective does not quantify the mistyped factor.

William H. Cabot coauthored the dynamic subgrid-scale model developed with Massimo Germano, Ugo Piomelli and Parviz Moin. Their 1990 report compared modeled unresolved stresses with data from direct numerical simulation and tested large eddy simulation of transitional and turbulent channel flow.^[ref] A participant retrospective identifies Cabot as a Stanford research associate during the collaboration.^[ref]

The numerical comparison illustrates an important issue in turbulence research: matching the objects being compared. A coarse simulation describes resolved motion, while a finely resolved reference contains additional fluctuations. The report filtered the reference data before some comparisons, so the comparison did not automatically count intentionally omitted small scales as errors in a resolved field.^[ref]

For an original arithmetic illustration, fine-grid velocities $0$ and $2$ have mean $1$ and mean square $2$. The square of their mean is only $1$. Replacing both samples by the mean preserves that mean but loses a fluctuation contribution of $1$. This elementary difference is why an averaged velocity and an averaged momentum flux contain different information, and why subgrid-scale stress requires explicit treatment.