The consistent themes in these research projects are the role of unsteadiness and the extent to which we can develop predictive models of turbulent motion in unsteady flows (as if it were not hard enough to do so in steady flows!).
Unsteady Laminar Flow Metering:
We have recently demonstrated a time-accurate flow rate sensor for unsteady
laminar liquid flows of Newtonian fluids in pipes. The principle on which this
sensor is based is a solution we have developed to the unsteady, fully-developed
Navier Stokes equations, that expresses the momentary flow rate as a
functional of pressure drop. The functional can be expressed as a term that
is a multiple of the momentary pressure drop (a Hagen-Poiseuille term) and a
convolution integral of pressure gradient history (an unsteady correction term).
The unsteady correction term, which is analytically exact, is necessary
because, in unsteady flows, the pressure drop is out of phase with the flow
rate. This unsteady flow meter comprises two micro-pressure sensors imbedded in
a long section of pipe, to measure the momentary pressure drop along the pipe,
and signal processing equipment to construct the convolution integral and deduce
the momentary flow rate. In test experiments, the principle appears to work
perfectly for duct flows with arbitrary unsteadiness histories. Dr. Brereton holds a
patent on this unsteady flow-metering technique.
Unsteady Turbulent Flow Metering:
In the unsteady flow studies described above, an exact solution is possible
because the unsteady, laminar fully-developed-flow equations are linear and
forward and inverse Laplace transformation, which facilitates the convolution
representation, is possible. In the corresponding
turbulent flow problem, it would be extremely useful to have an approximation to
an unsteady correction factor. However, if the same approach is to be taken as
for laminar flow, one must model the Reynolds stress in a closure which is at
most a linear function of velocity, velocity gradient or y, possibly with
an intrinsic time delay. We are presently evaluating the efficacy of such an
approach through comparisons with experimental measurements of momentary flow
rate and pressure gradient.
Unsteady Compressible Flow Metering:
One of the most challenging problems in automotive engine control is accurate
metering of air in intake flows, which are highly unsteady. Today's automobiles
typically use a hot-wire anemometer as the flow velocity sensor in a mass
airflow sensor. For engine control purposes, it is desirable to read a single,
steady voltage signal from the sensor that represents the integrated
contribution of a highly unsteady flow over each engine cycle. The use of an
instantaneous sensor without directional sensitivity in a bi-directional flow
complicates matters further. When deployed in nearly all today's automobile engines, a highly
damped voltage output from a potentially time accurate hot-wire sensor has to be
calibrated in situ by making a look-up table to provide the correct
measured average flow rate from voltage at fixed engine speeds. No calibration is made for
the infinite number of possible transient histories and so engine control
modules are typically programmed to disregard the sensor output during
transients. So much for making good use of a potentially valuable and accurate sensor! We
are presently exploring approaches to deducing the momentary flow rate in these
transient, compressible, almost frictionless duct flows by using simple sensors,
approximate solutions to the momentary equations of fluid motion, and smart
(real-time analog electronics) algorithms.
Unsteady Laminar Flow Control:
In the area of flow control, we have used exact solutions to the unsteady Navier
Stokes equations in fully-developed duct flows to find the optimal transients in
flow rate that will accelerate fluid from one value to another with the least
power input. In these problems, the power input is again a functional
of flow rate and the optimal solution departs significantly from the
quasi-steady one as less time is permitted for completion of the transient. We are
presently examining DNS databases to see if similar functional considerations
apply to optimization of turbulent flows.
Unsteady Turbulence Modeling:
The most difficult of the various areas of unsteady fluid mechanics we study is
that of developing turbulence models for use in unsteady turbulent flows. While
it is relatively easy to point to shortcomings of today's quasi-equilibrium
two-equation models in unsteady flows, it is another matter entirely to propose
better, more general models. We have explored some possibilities of using RDT
(rapid distortion theory) results as asymptotic limits towards which turbulence models should adjust at
high rates of strain, but real flows seem to reach only momentarily the
conditions under which RDT assumptions would apply. Moreover, we are presently
able only to compute RDT results in homogeneous turbulence, using the Reynolds-Kassinos
formulation, for arbitrary
strain-rate histories.
It seems as though non-local modeling in both time and space may be necessary to provide
adequate closures for Reynolds stress (and structural) descriptions of turbulence and research in this
general area is of continued, long-term interest.
Selected Publications:
Brereton, G. J. & Mankbadi, R. R.
Development of a rapid-distortion turbulence model for unsteady
hydrodynamic and scalar boundary layers.
Tenth Symposium on Turbulent Shear Flows, Penn State Univ.,
(1995).
Brereton,
G. J.,
The interdependence of friction, pressure gradient, and flow rate in
unsteady laminar parallel flow.
Phys. Fluids, 12, 3, 518, 2000.
Brereton, G. J. & Hwang, J-L.
The spacing of streaks in unsteady
turbulent wall-bounded flow.
Phys. Fluids A, 6 (7), 2446--2454, (1994).
Brereton, G. J. & Reynolds, W. C.
Dynamic response of boundary-layer turbulence to
oscillatory shear.
Phys. Fluids A, 3 (1), 178--187, (1991).
Brereton, G. J.
Stochastic estimation as a statistical tool for approximating turbulent
conditional averages.
Phys. Fluids A, 4 (9), 2046--2054, (1992).
Brereton, G. J. & Kodal, A.
An adaptive turbulence filter for decomposition of organized turbulent
flows.
Phys. Fluids 6 (5), 1775--1786, (1994).