Plasma Sources Science and Technology

ACCEPTED MANUSCRIPT

Post-stall Flow Control using Nanosecond Pulse Driven Dielectric Barrier

Discharge Plasma Actuators
To cite this article before publication: Jesse Little et al 2018 Plasma Sources Sci. Technol. in press https://doi.org/10.1088/1361-6595/aaf52f

Manuscript version: Accepted Manuscript

Accepted Manuscript is “the version of the article accepted for publication including all changes made as a result of the peer review process,
and which may also include the addition to the article by IOP Publishing of a header, an article ID, a cover sheet and/or an ‘Accepted
Manuscript’ watermark, but excluding any other editing, typesetting or other changes made by IOP Publishing and/or its licensors”

This Accepted Manuscript is © 2018 IOP Publishing Ltd.

During the embargo period (the 12 month period from the publication of the Version of Record of this article), the Accepted Manuscript is fully
protected by copyright and cannot be reused or reposted elsewhere.
As the Version of Record of this article is going to be / has been published on a subscription basis, this Accepted Manuscript is available for reuse
under a CC BY-NC-ND 3.0 licence after the 12 month embargo period.

After the embargo period, everyone is permitted to use copy and redistribute this article for non-commercial purposes only, provided that they
adhere to all the terms of the licence https://creativecommons.org/licences/by-nc-nd/3.0

Although reasonable endeavours have been taken to obtain all necessary permissions from third parties to include their copyrighted content
within this article, their full citation and copyright line may not be present in this Accepted Manuscript version. Before using any content from this
article, please refer to the Version of Record on IOPscience once published for full citation and copyright details, as permissions will likely be
required. All third party content is fully copyright protected, unless specifically stated otherwise in the figure caption in the Version of Record.

View the article online for updates and enhancements.

This content was downloaded from IP address 130.56.64.29 on 17/12/2018 at 13:44

Page 1 of 19 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1

1
2
3
4
5
Post-stall Flow Control using Nanosecond
6
7 Pulse Driven Dielectric Barrier Discharge
8

pt
9
10 Plasma Actuators.
11
12
13 Jesse Little,1 Ashish Singh,2 Timothy Ashcraft3 and Claudia Durasiewicz4

cri
14
15
16 Abstract
17 The efficacy of thermal perturbations generated by nanosecond pulse driven dielectric barrier discharge (ns-DBD)
18 plasma actuators for post stall flow control is explored on a NACA 0012 airfoil. Baseline and controlled flow
19 fields are studied using static pressure measurements, particle image velocimetry and constant temperature
20 anemometry. Experiments are primarily performed at Re = 0.74 x 106 (U∞ = 40 m/s) and α = 18°. Three regimes

us
21 of forcing are identified corresponding to separation control (0.92< F+ < 1.52), bluff body shedding (0.23 < F+ <
22 0.92) and an impulse-like response (F+ < 0.23). The response of the flow to a single high voltage pulse is also
23 examined and compared to other studies of transient separation control. Results show that the global structure of
24 the controlled flow is not specific to ns-DBD plasma actuator forcing. Rather, it is the local behavior of the ns-
25 DBD that results in control authority at conditions that are challenging for the majority of active flow control
26 actuators that rely on zero net mass flux momentum addition. The implications of these findings as well as open
27
28
29 1. Introduction
an
questions and suggestions for future work are discussed.

30
31 The use of plasma actuators for aerodynamic flow control gained considerable popularity in the early 2000s
32 driven primarily by the dielectric barrier discharge (DBD). Initially, DBDs driven by ac signals (ac-DBDs) were
dM
33 the dominant plasma actuator in the literature due to their low cost, simplicity and efficacy for controlling low
34 speed laboratory-scale flows. As these devices became more popular, researchers attempted to push ac-DBD flow
35 control demonstrations to higher speeds and were met with challenges. The ac-DBD control mechanism relies on
36 an electrohydrodynamic (EHD) effect which classifies it as a momentum-based actuator. While unique in its
37 method of momentum generation, it is still governed by relatively well-known amplitude scaling relationships
38 developed using synthetic jet actuators in the decades prior. Namely, the momentum required to control a given
39 flow scales with the freestream (or boundary layer edge) velocity or its square (Seifert and Tilmann 2009). This
40 means that progressively higher speed flows require actuators capable of producing more momentum. Estimates
41 on the capability of EHD devices for flow control were provided by Utkin et al. (2007). They suggest that control
42 authority can be expected up to approximately 100 m/s this has mostly held true to date although recent advances
using thick dielectrics and higher voltages are promising (Kelley et al. 2014).
43
pte

The desire to control higher speed flows (>100 m/s) motivates exploration of flow control actuators that
44
function via thermal energy deposition. Thermal perturbations can be generated using pulsed plasmas or lasers
45
and compelling demonstrations are widespread (Lazar et al. 2008; Roupassov et al. 2009; Samimy et al. 2010;
46 Little et al. 2012; DeBlauw et al. 2014; Kelley et al. 2014; Yugulis et al. 2014; Singhal et al. 2017; Webb and
47 Samimy 2017). The exact physics of how thermal perturbations couple to the local flow field remains unclear, but
48 the state of the art is summarized in Samimy et al. (2018) and Little (2018).
49 This work concerns the use of nanosecond pulse driven DBDs (ns-DBDs) for active flow control (AFC) of
50 the nominally separated flow on an airfoil in post-stall conditions. The use of a DBD arrangement offers
51
ce

considerable flexibility in the spatial distribution of the thermal perturbation at the expense of amplitude. In some
52 applications (e.g. airfoil leading edge), this has advantages over the localized arc filaments employed in high
53 speed jets and cavities (Samimy et al. 2010; Webb and Samimy 2017). The modern use of ns-DBD plasma
54 actuators for flow control was first examined by Starikovskii et al. (2008). Since then, researchers from around
55
56 1
Ac

57 Associate Professor, Department of Aerospace and Mechanical Engineering, University of Arizona

Corresponding Author, [email protected]
58 2
Graduate Student, Department of Aerospace and Mechanical Engineering, University of Arizona
59 3
Instructor, Department of Civil and Mechanical Engineering, United States Military Academy at West Point
60 4
Visiting Graduate Student, TU Berlin

1
 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1 Page 2 of 19

1
2
3 the globe have advanced the understanding of these device for AFC and other topics resulting in a special issue
4 on nonequilibrium nanosecond plasma aerodynamics to which this paper is a contribution.
5 Control of separation from the leading edge of an airfoil in stall remains the most popular platform for
6 demonstrations of ns-DBD plasma actuators. In this work, the capability of the ns-DBD to completely redefine
7 such a flow field is examined. Experiments are performed on a NACA 0012 airfoil, but findings are certainly not
8 specific to this model. Rather these results are intended to inform all efforts for controlling separation from an

pt
9 airfoil leading edge using thermal energy deposition. Experiments are conducted at Re=740,000 (U∞=40 m/s) and
10 primarily at 18 degrees incidence. The selection of 18 degrees is motivated, in part, by observations in the
11 literature. Wu et al., (1998) collected information from a variety of experimental and computational NACA 0012
12 studies. Various regimes were identified as functions of Reynolds number and angle of attack. The regimes were
13 classified according to the flow field behavior. This particular work focuses on the “low-frequency regime”, first

cri
14 identified experimentally by Zaman et al., (1989) at Re = 0.15 – 3.0 x 105 near static stall. They observed a low
15 frequency oscillation at a Strouhal number approximately one order of magnitude lower than the typical bluff
16 body shedding frequency (St=0.2). The low frequency oscillation was attributed to the transitional state of the
17 separating boundary/shear layer resulting in large unsteady forces. Wu et al., (1998) predicted that this low
18 frequency regime may extend up to Re of nearly 106 and 20 degrees incidence. The existence of this natural
oscillation suggests conditions are ripe for AFC especially using plasma actuators given their low profile and
19
minimal passive effect on the boundary layer. It is important to note that efficient use of AFC with any actuator
20

us
relies on the manipulation of natural flow instabilities. The fact that these instabilities are excited even in baseline
21
conditions implies they should be quite amenable to forcing provided a suitable actuator is employed.
22 The objective of this work is to demonstrate and explain the capability of ns-DBD plasma actuators to alter
23 the global flow structure over an airfoil in stall and this is relevant for thermal energy deposition-based actuators
24 in general. The following results also show the types of flow fields that can be realized using ns-DBD plasma
25 actuators. It is hoped that the reader will be inspired to define and pursue new avenues of research that are outside
26
27
28
29
of conventional post stall separation control studies.

2. Background
an
30 2.1 Leading Edge Airfoil Separation Control
31
32 In the arena of AFC, leading edge airfoil separation control has yielded noteworthy technological advances
dM
33 in recent decades. A widely accepted development is the use of pulsed blowing, suction or both as a more effective
34 technique than steady forcing for nominally two-dimensional excitation (Seifert et al. 1996). Related research
35 suggests that the range of dimensionless frequencies that are effective for separation control is on the order of
36 unity. This dimensionless frequency, often termed the reduced frequency, is defined as F+ = ffxsp/U∞, where ff is
37 the forcing frequency, xsp is the length of the separated region and U∞ is the freestream velocity (Darabi and
38 Wygnanski 2004b; Glezer et al. 2005). The characteristic length scale, xsp, is generally assumed to be the length
39 of the separated region over the body in question (Seifert et al. 1996). A reduced frequency of unity indicates that
40 one perturbation is excited during the propagation time of the freestream flow over the relevant length scale (i.e.
41 separated region). Selection of the actuator location is critical since the shear layer created between the freestream
42 and low speed region amplifies small perturbations introduced near the region of receptivity. In general, the
43 actuator location is chosen at or slightly upstream of the separation line to ensure the shear layer is excited by
pte

44 perturbations near its receptivity region. An indication of successful excitation is the formation of large spanwise
45 vortices developed via the shear layer (Kelvin-Helmholtz) instability. These vortices facilitate momentum transfer
between the freestream and separated region, thus reattaching flow to the body surface in the mean (Darabi and
46
Wygnanski 2004b; Melton et al. 2005). The vortices may couple with other instabilities downstream (separation
47
bubbles, wakes, etc.) and these dynamics are also considered in flow control efforts.
48
49 2.2 ns-DBD Plasma Actuators
50
51
ce

The use of plasma actuators for flow control is attractive for several reasons: relatively simple construction,
52 minimal interference with aero-lines on the controlled body, rapid time response and low power usage. In some
53 cases, they may also demonstrate high amplitude and bandwidth. The ns-DBD plasma actuator functions via
54 thermal energy deposition. The ns-DBD became popular after alternating current actuators (ac-DBDs), which
55 produce zero net mass flux (ZNMF) momentum-based perturbations, demonstrated limited effectiveness at
56 progressively higher flow speeds. This result was somewhat expected considering traditional amplitude scaling
Ac

57 via momentum coefficient, Cµ, or velocity ratio (Greenblatt and Wygnanski 2000). It is now known that the
58 momentum production of ac-DBDs can be increased by using thicker dielectrics and higher voltages (Corke et al.
59 2010). Knowledge of ns-DBDs has also expanded in recent years and it apparent that ns-DBDs possess some high
60 amplitude capability, but this effect is very much localized in space (Lehmann et al. 2016). Consequently, ns-

2
Page 3 of 19 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1

1
2
3 DBDs appear best suited for use in relatively thin boundary layers with well-known separation locations. The
4 leading edge of an airfoil post-stall is a prime example of this and ns-DBD control of such flows has been widely
5 reported (Starikovskii et al. 2009; Rethmel et al. 2011; Little et al. 2012; Kelley et al. 2014; Clifford et al. 2016;
6 Singhal et al. 2017).
7 The ns-DBD construction is identical to the ac-DBD and employs a high voltage waveform with very short
8 rise time. The input waveform consists of a pulse with rise time of a few tens of nanoseconds and voltages ranging

pt
9 between 5-50 kV. The extremely short rise time of the ns-DBD actuator generates rapid localized heating of near
10 surface gas by the plasma. The dielectric barrier and short pulse prevents arcing of the plasma and enables the
11 thermal effect to be spatially distributed along the device.
12
13 3. Experimental Facilities and Techniques

cri
14
15 3.1 Wind Tunnel
16
17 Experiments are performed in the Arizona Subsonic Wind Tunnel (ASWT). The ASWT is a closed loop
18 facility with a 0.9 m x 1.2 m x 3.65 m (3 ft x 4 ft x 12 ft) test section. The speed of the flow in the tunnel can be
19 varied up to U∞ = 80 m/s (262.467 ft/s). Mean flow uniformity is better than +/- 0.5% across the operating range.
20 Temperature in the tunnel is held within 0.44°C (1 °F) of a set point controlled by the operator using a heat

us
21 exchanger with chilled water supply. Over its operating range, the turbulence intensity is less than 0.15% for
22 flowpass = 10 kHz and less than 0.05% for fbandpass = 1 Hz – 10 kHz. Above the test section, a two-axis traverse is
23 used to mount probes for data acquisition. A pitot-static tube is mounted at the entrance of the test section to
acquire total and static pressures which are used to determine the flow speed and act as the reference pressure for
24
calculations of the pressure coefficient (CP).
25
26
27
28
29
3.2 Airfoil Model
an
The NACA 0012 airfoil used for this project was constructed using primarily polyoxymethylene (i.e., Delrin)
along with an aluminum trailing edge and steel spar at ¼ chord. The model has a chord length of c = 0.3048 m
30 (12 in) and a span of b = 0.8636 m (36 in). The surface of the model contains 64 static pressure taps. The suction
31 side of the airfoil at mid-span contains 23 taps and the pressure side at mid-span contains 13 taps. The taps are
32 more closely spaced near the leading edge to improve spatial resolution. The model also includes 28 taps at one-
dM
33 quarter and three-quarters span. These taps are used to evaluate three-dimensional effects in the flow.
34 The airfoil is mounted vertically in the test section using a steel spar which runs through the model at one-
35 quarter chord. Two plugs at the top and bottom hold the airfoil spar in place and two dowel pins, one on each
36 plug, ensure that the airfoil does not rotate independently thus eliminating twist. The plugs can be rotated to adjust
37 the angle of attack with uncertainty less than 0.25°. All experiments are performed at a chord-based Reynolds
38 number of 740,000 (40 m/s).
39
40 3.3 ns-DBD Plasma Actuators
41
42 Actuators are composed of two copper tape electrodes separated by Kapton tape dielectric arranged in an
43 asymmetric fashion. The two electrodes are placed immediately adjacent to one another to encourage uniform
pte

44 plasma generation. The discharge is directed upstream such that the exposed electrode does not disturb the
45 incoming boundary layer before encountering the plasma discharge. This counter-flow setup was shown to be
46 more effective than the co-flow arrangement (Bayoda et al., 2015a). The covered ground electrode is 12.7 mm
(1/2 in.) wide and the exposed high-voltage electrode is 6.35 mm (1/4 in.) wide. Each electrode has a thickness of
47
0.09 mm (3.5 mil). The dielectric barrier is composed of 2 layers of 0.076 mm (3 mil) Kapton tape with a dielectric
48
strength of 10 kV. Each layer of Kapton tape has a 0.04 mm (1.5 mil) layer of silicone adhesive such that the
49
actual Kapton thickness for each tape layer is only 0.04 mm (1.5 mil). The total thicknesses of the dielectric and
50 the device as a whole are 0.15 mm (6 mil) and 0.33 mm (13 mil), respectively. The actuator is applied to a 0.05
51
ce

mm (2 mil) base layer of kapton and then to the substrate in all cases. The actuator construction was optimized
52 based on previous work in airfoil separation control using ns-DBD plasma actuators (Little et al. 2012).
53
54
55
56
Ac

57
58
59
60

3
 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1 Page 4 of 19

1
2
3
4
5
6
7
8

pt
9
10
11
12 Figure 1. Asymmetric ns-DBD Plasma Actuator Schematic.
13

cri
High voltage nanosecond pulses are produced using a pulse generator designed and built by the Ohio State
14
University (OSU) Non-equilibrium Thermodynamics Laboratory (NETL). The pulser is magnetic compression
15
type, capable of both positive and negative polarity outputs with bandwidth of 10 kHz. Only positive polarity
16
pulses are studied here. Pulse voltage is controlled by a 5 kW Sorenson dc power supply capable of 0-700 V
17 output. Pulse parameters are strongly dependent on the DBD load. Input signals are generated by a Tektronix
18 AF6310 function generator. A Tektronix P6015A high voltage probe and custom current probe are used in
19 conjunction with a Tektronix TDS 2024C oscilloscope. Pulse voltage is measured at the output of the generator
20 while current is measured between the encapsulated electrode and ground. A pair of 40 cm transmission cables

us
21 connect the pulser to the DBD load.
22 Representative voltage and current traces on a 76.5-cm-long (30 in.) ns-DBD plasma actuator are shown in
23 Figure 2a. The ns-DBD pulse width is ~200 ns and reaches a peak voltage of ~7.5 kV with max current of ~0.7
24 A/cm. Figure 2b plots instantaneous power and energy coupled to the load at these conditions. Steady-state energy
25 of ~0.35 mJ/cm is realized approximately 150 ns after the beginning of the pulse. The energy is calculated by
26 integrating the absolute value of the product of voltage and current (Dawson and Little 2013). Due to the length
27
28
29
an
of the actuator, the peak power generated by the ns-DBD is only ~0.07 kW/cm. The pulse rise and fall time is
nearly 100 ns. The rise time and fall time is somewhat longer than previous experiments with this pulse generator
and the energy is somewhat lower (Little et al. 2012). This due to the relatively large size of the actuator. It will
30 later be shown that even this small energy is sufficient for exciting the flow.
31
32 (a) (b)
dM
33
34
35
36
37
38
39
40
41
42
43
pte

44
45
Figure 2. Sample traces of (a) Voltage and Current and (b) Instantaneous Power and Energy for a 76.5-
46 cm-long (30 in.) ns-DBD actuator.
47
48 The short duration pulses produced by ns-DBDs generate weak levels of momentum and are deemed
49 inconsequential in moderately high speed (M > 0.1) flow control demonstrations. Instead, the control authority
50 stems from rapid localized heating of the near surface gas (Joule heating). The signature of this heating is a
51
ce

compression wave shown in Figure 3. Figure 3a presents schlieren images visualized along the major axis of the
52 discharge similar to the arrangement in Figure 1. The cylindrical wavefront originates from the electrode interface
53 (actuator reference location in Figure 1) while the quasi-planar wave forms from spreading of the discharge on
54 the dielectric surface. The speed of the compression waves reaches sonic after only a short distance from the
55 surface, but is believed to be substantially greater very near the discharge (Roupassov et al. 2009; Takashima et
56 al. 2011). Measurements of overpressure indicate values in excess of a few hundred pascals (Bayoda et al., 2015b).
Ac

57 Figure 3b shows schlieren images viewed along the minor axis of the discharge. It is now apparent that the
58 cylindrical and quasi-planar waves are also composed of individual spherical wave fronts. These originate from
59 localized hot spots in the discharge that remain stationary from pulse to pulse for a given actuator. Measurements
60 of discharge overheat near the surface are challenging due to the spatial and temporal scales involved, but values

4
Page 5 of 19 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1

1
2
3 as high as 350-400K have been reported (Roupassov et al. 2009; Takashima et al. 2011). More recent studies by
4 Leonov et al. (2016) suggest that a secondary slow heating mechanism occurring ~100 µs after the pulse may be
5 influential. However, the relationship between the initial fast and slow heating mechanisms along with their
6 impact on flow control has yet to be fully explored. It should be noted that heating effects from ac-DBDs have
7 been disregarded in various studies (Enloe et al. 2004; Jukes et al. 2006; Sung et al. 2006).
8

pt
9 (a) (b)
10
11
12
13

cri
14
15
16
Figure 3: Sample phase averaged schlieren images of ns-DBD compression waves along the major (a) and
17
minor (b) axis of the discharge. The length scale (bottom left of image b) shows 2mm (Dawson and Little
18
2013).
19
20

us
21 3.4 Instrumentation
22
23 3.4.1 Pressure Measurements
24
25 A Scanivalve E-RAD400 is used to perform pressure measurements. This system consists of a ZOC33/64Px
26 electronic pressure scanning module used in conjunction with a RADBASE4000 base unit with a RAD 3200 A/D
27
28
29
an
converter. The pressure range of the selected module is pmax = 6,895 Pa (1.0 psi). The resolution of the system is
given as ±0.12% of the full-scale reading, or presolution = 8.32 Pa (0.0012 psi). The freestream dynamic pressure at
U∞ = 40 m/s is q = 980 Pa (0.142 psi), meaning that resolution of the system is 0.8% of dynamic pressure. Fifty
30 pressure data samples are collected in 32 s (fsample = 1.56 Hz). The resulting values are used to compute the static
31 Cp distribution over the airfoil surface having a statistical uncertainty on Cp of at most 2% with a confidence
interval of 95%.
32
dM
33
3.4.2 Particle Image Velocimetry
34
35 A LaVision 2D PIV system (DaVis 8.3.0) is used to obtain spatially resolved velocity data. Submicron DEHS
36 seed particles are injected into the tunnel using a LaVision aerosol generator. A double-pulsed 200 mJ Quantel
37 Evergreen laser is used in conjunction with spherical and cylindrical lenses to form a light sheet for the PIV
38 measurements. The laser repetition rate is nominally 15 Hz. The time separation between laser pulses used for
39 particle scattering is set according to the flow velocity, camera magnification, and correlation window size. Images
40 corresponding to the pulses from each laser head are acquired by 16 bit 5.5 megapixel LaVision Imager sCMOS
41 cameras incorporating Nikon Nikkor 50 mm f 1.2 lenses and narrow bandpass optical filters. For each image pair,
42 sub-regions are cross-correlated using decreasing window size (642–322 pixel2) multi-pass processing with 50%
43
pte

overlap. The resulting velocity fields are post-processed to remove vectors with correlation peak ratio less than
44 1.5 and a correlation coefficient below 0.5. Removed vectors are replaced using an interpolation scheme based on
45 the average of neighboring vectors. A 3 x 3 Gaussian smoothing filter is also applied to the calculated velocity
46 fields. The spatial resolution of the PIV is at or better than 3.62 mm (1.2% of chord) in all cases. Time-averaged
47 data are produced from 500 instantaneous samples resulting in an uncertainty on velocity of 5.62% or less of
48 freestream using a 95% confidence interval. Phase-averaged data are composed of 250 samples per phase,
49 rendering uncertainties less than 1.5% of freestream using a 95% confidence interval.
50
51 3.4.3 Constant Temperature Anemometry
ce

52
53 Constant temperature anemometry (CTA) is performed using a Dantec Streamware Pro system with 55P01
54 probes and 55H21 supports. Probes are positioned in the wake of the disturbance airfoil at x/c=6 at various
spanwise locations. The sampling rate of the hot wire measurements is 10 kHz for 200,000 total samples resulting
55
in a total measurement duration of 20 s. The power spectrum density (PSD) is calculated using 60 blocks each
56
containing 8192 samples. A Hanning window is applied to each block and an overlap of 50% is employed. The
Ac

57
magnitude squared coherence is employed to evaluate the two-dimensionality in the wake using the same
58 parameters as the PSD.
59
60

5
 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1 Page 6 of 19

1
2
3
4 4. Results
5
6 4.1 Test Conditions
7
8 A NACA 0012 airfoil at Re = 0.74 x 10 6 (U∞ = 40 m/s) is the test model. Experiments are conducted at

pt
9 atmospheric temperature and pressure. The aerodynamic characteristics of the baseline model are shown in Figure
10 4. Experimental data for lift coefficient is calculated from the surface pressure distribution. Theoretical predictions
11 using thin airfoil theory and XFLR5 are also included in Figure 4. The increasing and decreasing C L curves
12 represent sweeps done by going from low to high angle of attack and vice versa, respectively. A well-known
13 hysteresis behavior is observed near and after the stall angle. Variations between the two experimental data sets

cri
14 in the linear CL-α regime provide an indication of the experimental uncertainty on angle of attack. Overall, the
15 sectional lift coefficient matches thin airfoil theory in the linear region of the curve. XFLR5 predictions are also
16 in good agreement up to an angle of attack of 11-12 degrees. Both experimental data sets show excellent agreement
17 above 15 degrees. The post-stall region, specifically 18 degrees which is outside of hysteresis effects, is the focus
18 of this work. For all measurements, the baseline separated flow condition is established before initiating ns-DBD
19 plasma actuator.
20

us
21
22
23
24
25
26
27
28
29
an
30
31
32
dM
33
34
Figure 4. NACA 0012 lift coefficient as a function of angle of attack at Re=740,000 (40 m/s).
35
36
37 4.2 Pressure Measurements
38
39 To determine optimal placement of ns-DBD plasma actuators, Cp measurements were acquired for various
40 locations and frequencies. The actuator locations are defined as an arc length from the leading edge (x/c = 0) to
41 the plasma actuator reference location on the suction surface (see Figure 1). The locations tested include 1 mm, 2
42 mm, 3 mm, 4.5 mm, and 6 mm. Frequency sweeps were acquired at each location in the range F+ = 0.08 – 7.62
43 (ff = 10 Hz – 1 kHz) at angles of 14° (CL,max), 17°, 18°, and 19°. The forcing amplitude is kept constant in all cases
pte

44 at 0.35 mJ/cm unless stated otherwise. Installation of the actuator prevents the use of pressure taps at the leading
45 edge. On the pressure side, seven mid-span and two 3D pressure taps are lost while on the suction side, five mid-
46 span and two 3D pressure taps are lost. Thus, usable data from the pressure tap nearest the leading edge on the
47 suction side is at x/c = 0.09. This investigation found 3 mm as the optimal location for AFC at α = 18° since it
48 yielded the lowest Cp. It is also important to note that the actuator is applied such that the discharge is directed
49 upstream. This means that the boundary layer developing from the stagnation line encounters plasma before seeing
50 any discontinuities due to the exposed electrode.
51 Frequency sweeps for the 3 mm actuator location at two angles of attack (16° and 18°) are shown in Figure
ce

52 5. At 18°, F+ ≈ 1.14 corresponds to the lowest CP at x/c = 0.09. This will later be shown as the best condition for
separation control. The shape of the min(Cp) curve implies that some optimal forcing frequency is established
53
and this indicates that coherent structures are being produced, which will later be shown using PIV. A comparison
54
with the 16° case emphasizes fundamental differences in the flow response with angle of attack. At the lower
55
angle of attack, the CP value decreases significantly with increasing forcing frequency. After F+ ≈ 0.5, the CP
56 value is essentially constant. The trend indicates that in this condition, AFC acts as an active boundary layer trip
Ac

57 if applied above a certain frequency. Similar behavior has also been observed for a NASA EET airfoil (Little et
58 al. 2012). At higher angles, boundary layer transition alone is not sufficient for reattaching the nominally separated
59 flow. These findings are applicable to AFC with ns-DBDs on airfoils in general.
60

6
Page 7 of 19 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1

1
2
3 Observations at both angles are consistent with data in Wu et al., (1998). At 16°, increasing Re to 10,000,000
4 enters the turbulent, steady attached flow regime and this is approximated by the active boundary layer trip here.
5 At 18°, turbulent steady attached flow is unexpected even at Re=10,000,000. Instead, turbulent steady separated
6 flow is predicted. Consequently, fruitful attempts at separation control at this condition should leverage other
7 means such as excitation of coherent structures. This emphasizes an important point about AFC which should be
8 considered for any new actuation technique. It is the flow field that determines the response to actuation and any

pt
9 attempt to execute a flow control strategy should be firmly grounded in flow physics. Ns-DBDs are a somewhat
10 novel device for flow control since they operate via thermal energy deposition and this contrasts with most flow
11 control actuators that rely on momentum addition included ac-DBDs. However, even this unique type of
12 perturbation is still governed by the flow field.
13

cri
14
15
16
17
18
19
20

us
21
22
23
24
25
26
27
28
29
an
Figure 5. Suction side CP at x/c = 0.09 vs. dimensionless forcing frequency, F+, for NACA 0012 airfoil at Re
= 0.74x106 and α = 16° & 18°.
30
31
32 Observations from a single pressure tap on the suction surface are verified using surface pressure data for the
dM
33 other available sensors. In addition, the spanwise uniformity of flow is investigated using pressure taps at ¼ and
34 ¾ span. Figure 6a-c shows Cp plots at mid-span and two the adjacent spanwise locations for three actuation
35 frequencies. It will later be shown that these frequencies correspond to three forcing regimes. The missing data
36 near the leading edge is due to the actuator’s presence. Only subtle variations in Cp are observed at off-midspan
locations, hence the flow is two dimensional over at least the central ½ span of the airfoil during application of
37
AFC cases. Figure 6d shows the select AFC cases along with the baseline. The baseline flow has a near-zero
38
pressure gradient along the suction side of the airfoil, indicating deep stall (as expected). All forcing frequencies
39
show an improvement over the baseline and the best case for separation control is F+ ≈ 1.14 (ff = 150 Hz) as
40 observed in Figure 5. Note that forcing at F+ ≈ 1.52 (ff = 200 Hz) is also quite successful, but reaches a slightly
41 lower CP value. These cases form the basis of further investigations using PIV and CTA.
42
43
pte

44
45
46
47
48
49
50
51
ce

52
53
54
55
56
Ac

57
58
59
60

7
 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1 Page 8 of 19

1
2
3
4
5
6
7
8

pt
9
10
11
12
13

cri
14
15
16
17
(a) (b)
18
19
20

us
21
22
23
24
25
26
27
28
29
an
30
31
32
dM
33 (c) (d)
34 Figure 6. (a) Mid-span and 3D Cp distribution for F+ ≈ 0.15 (ff = 20 Hz), (b) F+ ≈ 0.46 (ff = 60 Hz), (c) F+ ≈
35 1.14 (ff = 150 Hz) and (d) Cp distribution for various forcing frequencies.
36
37
38 4.3 PIV Results
39
40 2D PIV is used to map the flow field from the leading edge to approximately 4.5 chord-lengths downstream
41 in the airfoil wake. A single camera captures flow over the airfoil surface, with a mask chosen to omit regions
42 obscured by the wall plug. The wake (from x/c = 1 to ≈ 4.5) is captured in a subsequent and identical test run
43 using two cameras side-by-side. The images are then stitched together since they are either time-averaged or
pte

44 phase-averaged relative to the actuation signal. This configuration provides a reasonably large field of view to
45 study the ns-DBD’s ability for controlling flow separation and producing coherent structures in the wake. All PIV
46 measurements are taken at Re = 0.74 x 106 (40 m/s) with the airfoil in deep stall at α = 18°.
PIV data are acquired for the baseline and three other frequencies that correspond to different regimes of flow
47
response. The lowest frequency (F+ ≈ 0.15 (ff = 20 Hz)) corresponds to a case in which the flow responds as if
48
perturbed by an impulse-like signal. Moderate forcing at F+≈ 0.46 (ff = 60 Hz) sets up a very coherent oscillation
49
in the wake while F+ ≈ 1.14 (ff = 150 Hz) is optimal for separation control. The motivation for identifying these
50 specific cases will be discussed in later sections.
51
ce

Figure 7 shows dimensionless time-averaged streamwise velocity 𝑈 ̅ ∗ contours for the baseline and forcing
52 cases, along with the corresponding velocity profiles at x/c = 2. The color scale is kept constant throughout.
53 Clearly, the airfoil is fully-stalled in the baseline case. On forcing, flow separation visibly reduces with increasing
54 frequency and this matches the CP results in Figures 5 and 6. Quantitatively, this reduces the momentum deficit
55 in the wake, as seen in the corresponding velocity profiles at x/c = 2. These profiles show a progressive decrease
56 in wake size, from the baseline to the best case for separation control (F+ ≈ 1.14 (ff = 150 Hz)).
Ac

57
58
59
60

8
Page 9 of 19 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1

1
2
3
4
5
6
7
8

pt
9
10
11
12
13

cri
14
15
16
17
18
19
20

us
21
22
23
24
25
26
27
28
29
an
Figure 7. Plots of time-averaged dimensionless streamwise velocity at Re = 0.74x106, α = 18° (a) Baseline
case (no control) and ns-DBD forcing at (b) F+ ≈ 0.15 (ff = 20 Hz) (c) F+ ≈ 0.46 (ff = 60 Hz) and (d) F+ ≈ 1.14
30
(ff = 150 Hz).
31
32
dM
33 The reduction in wake size is related to the actuator’s ability to produce disturbances in the flow. This is most
34 strikingly seen in the phase-averaged transverse velocity fluctations (Figure 8), where the flow is structured as
35 pairs of positive and negative velocity regions. The figure shows four equally distributed phases corresponding to
36 the three forcing cases. The first phase (top row in Figure 8) is acquired just after the high voltage pulse is applied
37 and the subsequent phases are separated by π/2. The flow response appears nearly sinusoidal for F+ = 0.46 (60
38 Hz) and also at F+ = 1.14 (150 Hz) over the model. In the low frequency case (F+ = 0.15 (20 Hz)), the behavior
39 is impulse-like in that shortly after actuation a large structure develops and propagates over the airfoil into the
40 wake followed by a quiet period with no discernable structure.
41 The phase averaged velocity fluctuations differ in magnitude and structure between all forcing cases. In the
42 low frequency case (F+ = 0.15 (20 Hz)), the scale of this structure is approximately the length of the airfoil chord.
43 After it propagates into the wake, a small positive velocity fluctuation is observed over the airfoil as the flow
pte

44 begins relaxing back to its baseline state, but doesn’t quite achieve it before the next plasma pulse. At F+ = 0.46
45 (60 Hz), a very organized pattern of positive and negative fluctuations is visible in the wake and throughout the
46 entire field of view for all phases. Their amplitude appears strong near the airfoil, but is only about 25% of the
47 freestream at the downstream end of the field of view. At F+=1.14 (150 Hz), the fluctuations in the wake are much
48 weaker compared to the two lower frequency cases. However, these organized coherent structures dominate the
49 flow over the model creating the best conditions for separation control. Forcing at this frequency creates a strong
50 suction peak near the leading edge and reattaches the flow in the mean over nearly the entire suction surface (see
Figure 6). The effectiveness of coherent structures in entraining freestream momentum and reattaching separated
51
ce

flows is well established (Greenblatt and Wygnanski 2000). Consistent with previous studies, the control
52
mechanism in this case is not due to flow tripping, but rather due to the excitation of flow instabilities through ns-
53
DBD thermal perturbations.
54
55
56
Ac

57
58
59
60

9
 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1 Page 10 of 19

1
2
3
4
5
6
7
8

pt
9
10
11
12
13

cri
14
15
16
17
18
19
20

us
21 Figure 8. Phase-averaged transverse velocity fluctuations for ns-DBD forcing at F+ ≈ 0.15 (20 Hz), F+ ≈ 0.46
22 (60 Hz), F+ ≈ 1.14 (150 Hz).
23
24 Figure 9 shows the vorticity dymanics associated with these disturbances. At the two lower frequencies (F+
25 = 0.15 (20 Hz) & F+ = 0.46 (60 Hz)), vortex shedding from the pressure side is dominant in the wake and the
26 clockwise rotating structure shedding from the suction surface is weaker. At F+=1.14 (150 Hz), the vorticity over
27
28
29
an
the airfoil is confined to a more narrower region (due to flow reattachment), while in the wake the vorticity is
diffuse with no apparent structure. These results reflect the dynamics seen in the phase averaged transverse
velocity fluctuations (Figure 8). However, certain features like the coherent structures over the airfoil at F+=1.14
30 (150 Hz) are obscured when only looking at vorticity.
31
32
dM
33
34
35
36
37
38
39
40
41
42
43
pte

44
45
46
47
48
49
50 Figure 9. Phase-averaged vorticity for ns-DBD forcing at F+ ≈ 0.15 (20 Hz), F+ ≈ 0.46 (60 Hz), F+ ≈ 1.14 (150
51 Hz). Note that the unphysical vertical lines near x/c = 3 in the vorticity fields are due to stitching of the two
ce

camera images.
52
53
The PIV data in Figures 8 and 9 are acquired near the centerline of the model and the two dimensionality of
54
the flow is not apparent. Certainly, the surface pressure data in Figure 6 suggests that nominally 2D behavior is
55 occuring over at least the central ½ of the suction surface of the airfoil when AFC is applied. It will later be shown
56 that a signifigant portion of the wake also possesses 2D features at some forcing conditions. This means that in
Ac

57 these conditions the ns-DBD is producing a very uniform 2D disturbance. Ns-DBDs are known to produce small
58 hot spots along the span as described in Section 3.3, but for the frequencies and amplitudes considered here this
59 does not manifest as a three-dimensional disturbance.
60

10
Page 11 of 19 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1

1
2
3 4.4 Constant Temperature Anemometry
4
5 CTA is employed to quantify the spectra and spanwise coherence in the excited wake. Data are collected over
6 a range of forcing cases from F+ = 0.08 – 1.22 (ff = 10 – 160 Hz) in increments of ΔF+ = 0.08 (Δff = 10 Hz). To
7 match PIV measurements, focus is placed on F+ = 0.15, 0.46, and 1.14 (ff = 20, 60, and 150 Hz respectively). All
8 CTA data are acquired at the ¼ chord with x/c = 6 and varying spanwise locations. Note that ¼ chord is also the

pt
9 axis of airfoil rotation.
10 Figure 10 shows PSD at mid-span for the baseline and three forcing cases. The transverse location of all CTA
11 data is at ¼ chord. All cases present a well-developed inertial subrange over more than a decade of St. The
12 frequency at which the inertial subrange becomes apparent is not appreciably different from the various forcing
13 cases. The baseline case lacks a clear spectral peak as expected. The same is true of the highest forcing frequency

cri
14 case (F+ ≈ 1.14). Recall that this forcing frequency results in the best control of separation. PIV data indicate a
15 robust pattern over the airfoil (see Figures 8 & 9), but this behavior quickly diminishes in the wake such that the
16 forcing signal is absent at x/c = 6. Note that PSD values at high frequency are slightly reduced in this case. Forcing
17 at F+ = 0.46 (60 Hz) results in a single dominant peak in the spectrum while lower frequency forcing (F+=0.15
18 (20 Hz)) produces a peak at the fundamental (St=0.15 (20 Hz) and two harmonics (St=0.3 (40 Hz) and (St=0.45
(60 Hz)). All forcing cases have higher broadband PSD magnitude compared to the baseline below the onset of
19
the inertial subrange (St < 0.45 (60 Hz)).
20

us
21
22 (a) (b)
23
24
25
26
27
28
29
an
30
31
32
dM
33
34 Figure 10. (a) Mid-span PSD and (b) coherence spectra across the central ½ of the wake (1.5c) for select
35 forcing frequencies at x/c=6.
36
37
38 Figure 10b shows the coherence between probes mounted at ¼ and ¾ span (separation of 1.5c) at x/c=6. As
39 with PSD, the baseline and high frequency cases (F+=1.14 (150 Hz)) have no clear peaks. Forcing at F+=0.46 (60
40 Hz) results in high coherence across the central 1.5c of the wake, with a normalized value of approximately 0.9.
41 In the low frequency case (F+=0.46 (20 Hz)), multiple harmonics exist and are consistent with the PSD. Overall,
42 Figure 10 shows that forcing at F+ = 0.15 (20 Hz) and F+ = 0.46 (60 Hz) excite the wake with a high degree of
43 coherence over the central ½ span of the wind tunnel (1.5c).
pte

44 This last observation is strengthened by contour plots of PSD and coherence for the entire range of forcing
45 frequencies (Figure 11). The dashed line indicates a slope of one. Note that gaps in the high PSD and coherence
46 levels are solely due to limited resolution in the tested forcing frequencies (the PSD data are acquired at higher
47 resolution). Essentially any frequency can be excited in the wake below the inertial subrange (St<0.92 (120 Hz)).
48 This is most easily seen in the coherence plots (Figure 11b), since the PSD peaks at high St are indistinguishable
49 in the color scale (Figure 11a). Forcing below F+=0.3 (40 Hz), gives a peak at the fundamental and multiple
harmonics and these are also quite coherent across the central ½ of the tunnel (1.5c). This observation is consistent
50
with the subset shown in Figure 10 as well as PIV data in Figures 8 and 9. The existence of harmonics indicates
51
ce

an impulse-like disturbance. Forcing between 0.3<F+<0.92 (40<f<120 Hz) produces a single dominant frequency
52
in the wake (also in agreement with Figures 8 & 9) and is quite coherent across the central ½ span (1.5c). In this
53 case, forcing produces a coupling between instabilities in the shear layer and the wake. More specifically, ns-
54 DBD induced disturbances at the leading edge are amplified by the shear layer instability over the model chord.
55 The frequencies in question (0.3<F+<0.92) are also in the range of typical bluff body shedding frequencies
56 (St≈0.2) when scaled by the projection of the chord in the streamwise direction (0.094 m). Higher frequencies
Ac

57 (F+>0.92 (120 Hz)) are useful for separation control, but do not produce significant disturbances at x/c=6 as
58 previously discussed.
59
60

11
 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1 Page 12 of 19

1
2
3
4 (a) (b)
5
6
7
8

pt
9
10
11
12
13

cri
14
15
16
17
18 Figure 11. Contour plots of (a) PSD and (b) coherence spectra across the central ½ of the wake (1.5c) for
19 select forcing frequencies at x/c=6. The dashed line indicates a slope of one.
20

us
21
4.5 Transient Behavior
22
23 The previous sections examine repetitive forcing on the flow field. The interaction of single pulse with the
24 flow is a fundamentally interesting question and sheds light on how the global structure evolves from local changes
25 at the leading edge due to actuation. To this aim, the transient response of separated flow to a single ns-DBD pulse
26
27
28
29
an
is examined. Apart from fundamental research, studies of so-called transient separation control have practical
merit in that they define the minimum response time of flow to actuation. In addition, they represent a limiting
case in which a single low cost (in terms of energy consumption) disturbance is used to control the flow. Studies
of transient separation control have been undertaken using a variety of fluidic actuators (Amitay and Glezer 2002;
30 Darabi and Wygnanski 2004b; Darabi and Wygnanski 2004a; Amitay and Glezer 2006; Mathis et al. 2007; Mathis
31 et al. 2009; Williams et al. 2009; Brzozowski et al. 2010; Albrecht et al. 2015), but it remains relatively unexplored
32 with thermal energy deposition.
dM
33 A single actuation pulse (Q = 0.23 mJ/cm) is applied to the separated shear layer over the airfoil. The flow
34 response is visualized using the vorticity field in Figure 12 at various convective times. The baseline flow is also
35 included for reference. Shortly after the actuation pulse at t/Tconv = 0, the shear layer severs at the leading edge
36 (t/Tconv = 0.2). At the point of severance, a clockwise rotating vortex is created and the freestream carries it
37 downstream, as seen in subsequent time delays. This shear layer vortex expands and weakens as it advects over
38 the chord. Finally, it moves into the wake and out of the field of view. Another structure is observed near the
39 leading edge of the airfoil surface just after the shear layer is severed. This second vortex grows over the suction
40 surface of the airfoil until the separated shear layer is fully reformed (t/Tconv =10).
41 The ns-DBD actuation pulse is much shorter than the convective time scale (Tpulse/Tconv= 2.4x10-8) yet the
flow remains at least partially attached for up to two convective times. These results are qualitatively similar to
42
other transient separation control studies which use fluidic actuators, where reattachment persists for several
43
pte

convective times (Brzozowski et al. 2010). It should be noted that the employed ns-DBD pulse duration, Tpulse=200
44
ns, is quantified based only on electrical data (e.g. Figure 2). Defining the precise time-scale of the thermal
45 disturbance remains unclear, but the reported value is certainly an underestimate. Regardless, ns-DBD plasma
46 forcing occurs on a time scale that is substantially shorter than fluidic actuators, including pulsed combustion jet
47 actuators having (Tpulse/Tconv=0.05) to which these results are comparable (Brzozowski et al. 2010). However, in
48 agreement with Brzozowski et al. (2010) the effect of this very short duration actuator pulse persists for up to 8-
49 10 convective times. The transient separation due shedding of the second vortex occurs in a manner that is
50 consistent with dynamic stall.
51
ce

52
53
54
55
56
Ac

57
58
59
60

12
Page 13 of 19 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1

1
2
3
4
5
6
7
8

pt
9
10
11
12
13

cri
14
15
16
17
18
19
20

us
21
22
23
24
25
26
27
28
29
an
Figure 12. Response of the separated shear layer to ns-DBD forcing at various convective times.
30
31
32 A control volume analysis of the circulation is computed over the area shown in Figure 13a to provide a more
dM
quantitative assessment of the transient response to thermal energy deposition. This location provides context to
33
similar studies in literature and specifically data from Brzozowski et al. (2010). Brzozowski et al. (2010) used a
34
similar window size, but just downstream of the airfoil trailing edge such that vorticity flux from both the suction
35
and pressure side was included. The latter had very little influence on the calculation, hence our results provide a
36 reasonable comparison (although located further upstream).
37 𝑑Γ
Figure 13b shows the variation in normalized vorticity flux ( = ∫ 𝑢𝜔 𝑑𝐴) with convective time. Similar
38 𝑑𝑡
39 fluctuations in circulation are observed in both data sets. The first local maximum (t/Tconv < 1) corresponds to the
40 location where the severed vortex passes the integration domain (see Figure 11). The minimum value of dΓ/dt
41 near t/Tconv=1 is due to an absence of vorticity in the severed region behind the vortex, and this appears more
42 extreme in the ns-DBD case. Vorticity flux is maximum at t/Tconv ≈ 2, where the second vortex crosses the
43 integration domain and again the ns-DBD value is larger. Both data sets relax to zero after approximately three
pte

convective times.
44
Figure 13c shows the change in circulation relative to the baseline which is calculated by integrating dΓ/dt.
45
The behavior is similar, but the ns-DBD lacks the initial local maximum at t/Tconv < 1. The minimum at t/Tconv = 1
46
is nearly identical while the subsequent values vary between the data sets. In both cases, circulation eventually
47 drops to its baseline value in about 10 convective times as qualitatively observed in Figure 12. The modest peak
48 near t/Tconv=7 in the ns-DBD case occurs when the separated flow reforms over the airfoil. This peak may be a
49 by-product of insufficient spatial resolution in the PIV setup which is highlighted by the diffuse nature of the
50 vorticity in the final stages of this process. As such, it is unclear if this is physically significant or just a
51
ce

representation of the uncertainty on the calculation. Apart from the differing window locations between the
52 studies, Brzozowski et al. (2010) used a different airfoil (NACA 4415), actuator (pulsed combustion jet), location
53 (x/c=0.2) and Reynolds number (Re = 570,000). Despite these significant differences, the global flow response is
54 quite similar and these results also coincide with momentum-based forcing using ac-DBDs (Durasiewicz et al.
55 2018).
56 In summary, these findings show that the global response of shear layers to single pulse actuation on stalled
Ac

57 airfoils is dramatically consistent despite differences in Reynolds number, geometry, control mechanism and
58 actuator location. The very different control mechanisms employed in this work and others influence the local
59 behavior. However, the global flow response is strikingly similar. This observation is consistent with findings of
60 Albrecht et al. (2015) who used various Lorentz force actuator configurations, but with much longer time scales

13
 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1 Page 14 of 19

1
2
3 (Tpulse/Tconv = 0.5) and observed similar global flow response in each case. Collectively, these results imply that
4 the flow itself (or the flow instability in this case) governs the response and this is mostly independent of the
5 actuator type. The uniqueness of ns-DBDs and thermal energy deposition in general lies in their ability to control
6 such flows at high speeds often in excess of ZNMF momentum-based devices.
7
8

pt
9 (a)
10
11
12
13

cri
14
15
16
17
18
19
20 (b) (c)

us
21
22
23
24
25
26
27
28
29
an
30
31 Figure 13. (a) Reference window used for calculations (b) Variation in normalized vorticity flux (−
𝒅𝜞 𝟏
)
𝒅𝒕 𝑼𝟐
32 𝜟𝜞∗
∞
dM
33 with time (c) Variation in normalized circulation (−
𝜞𝟎
) with time.
34
35 Finally, the influence of forcing amplitude on ns-DBD generated disturbances is presented. The available
36 pulse energy range is limited to a factor of three because of the actuator length. It is clear from Figure 14 that
37 variation in pulse energy (in this range) produces no observable change in the flow structure. Contours of both
38 vorticity and transverse velocity are nearly identical. It is plausible that a further increase in pulse amplitude would
39 result in a change in the flow response. Indeed, a reduction in control authority above a certain pulse energy has
40 been observed for the reataching shear layer downstream of a backward facing step (Akins et al. 2015). Recently,
41 Komuro et al. (2018) used a much smaller airfoil (Re=260,000) to show variations in the transient flow response
42 in the range of 0.006-0.6 mJ/cm/pulse. A slight dropoff in L/D and CL was observed above a certain energy level.
43 It should be feasible to replicate this study on larger models given an adequate power supply.
pte

44
45
46
47
48
49
50
51
ce

52
53
54
55
56
Ac

57
58
59
60

14
Page 15 of 19 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1

1
2
3
4
5
6
7
8

pt
9
10
11
12
13

cri
14
15
16
17
18
19
20

us
21
22
23
24
25
26
27
28
29
an
30
31
32
dM
Figure 14. Normalized vorticity and transverse velocity for pulse energies of 0.12, 0.23 and 0.35 mJ/cm at
33 t/Tconv = 0.2
34
35
36 5. Discussion and Conclusions
37
38 Post-stall flow control has been demonstrated on a NACA 0012 airfoil with an aspect ratio of three at Re =
39 740,000 (40 m/s) using ns-DBD plasma-generated thermal perturbations. Nominally 2D actuation is applied near
40 the leading edge with a pulse energy of 0.35 mJ/cm. For modest post-stall angles of attack (e.g. 16 degrees), ns-
41 DBD forcing at F+>0.5 reattaches the nominally separated flow by promoting boundary layer transition. This is
42 evident from a strong suction peak near the leading edge and the lack of a preferred forcing frequency. These
43
pte

findings are consistent with demonstrations on other airfoils in the literature (e.g. Little et al. (2012)) and
44 numerical predictions for the NACA 0012 (Wu et al. 1998) which suggest that transition should enable flow
45 reattachment at this angle. Increasing the angle of attack to 18 degrees removes boundary layer transition as a
46 viable separation control strategy, necessitating leverage of the shear layer instability for AFC.
47 At frequencies near F+=1, ns-DBD generated perturbations create large coherent structures over the chord of
48 the airfoil. These structures entrain freestream momentum and reattach flow in the mean. The physics of this
49 process are governed by the instability of the separated shear layer and the flow response is consistent with that
50 observed using various flow control actuators that operate via momentum addition (e.g. synthetic jets and ac-
51 DBDs) or even mechanical displacement (e.g. vibrating ribbon) (Greenblatt and Wygnanski 2000). The signature
ce

52 of these structures is quite apparent over the model chord and appears to be nearly two-dimensional over the
central ½ span based on surface pressure.
53
Reducing the forcing frequency (0.3<F+<0.92 (40<f<120 Hz)) enables a coupling between the shear layer
54
and wake instability that results in large sinusoidal oscillations characterized by a single peak in the PSD. In
55 general, the wake can be excited at any frequency (at least downstream to x/c=6 and somewhat beyond) provided
56 it is near or below the onset of the inertial subrange. In this case, behavior is reminiscent of bluff body shedding.
Ac

57 which is expected near St=0.2 (Zaman et al. 1989). If one considers the relevant length scale as the projection of
58 the airfoil chord in the streamwise direction, then these dimensionless forcing frequencies (0.3<F+<0.92) decrease
59 by a factor of ~3 and are comparable to bluff body shedding expectations (0.1<St<0.3). The coherence of ns-DBD
60 generated disturbances is found to reach levels well-above 0.5 at x/c=6 across the central half of the wake

15
 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1 Page 16 of 19

1
2
3 suggesting some level of 2D behavior. This does not preclude the existence of a secondary instability (spanwise
4 undulations of the wake) and additional measurements are required to investigate this.
5 Low frequency forcing (F+ < 0.15 (20 Hz)) generates a single vortical disturbance after which the flow relaxes
6 back to something resembling the baseline state. The full re-establishment of the baseline separated flow condition
7 is dependent on the frequency of forcing. More specifically, very low frequencies (e.g., F+ ≤ 0.0375 (5 Hz))
8 present as a repetitive impulse with multiple higher harmonics in the PSD measured at x/c=6 from the leading

pt
9 edge. Note that the impulse response produced by thermal energy deposition using localized arcs has been
10 explored in supersonic jets (Sinha et al. 2012). Inspection of phase-averaged transverse velocity fluctuations and
11 vorticity show that a negative region of vorticity is formed by severing the separated shear layer with each
12 actuation pulse. The response of the flow to a single ns-DBD pulse is strikingly similar to other published transient
13 separation control studies using different airfoils, actuators and forcing locations. This highlights the robustness

cri
14 of the flow physics (in this case the flow instability) that governs response. A single pulse of actuation severs the
15 separated shear layer resulting in a clockwise vortex being shed into the wake. The separated flow then gradually
16 re-establishes over approximately 10 convective times in a manner that resembles dynamic stall. Varying the
17 amplitude of forcing by a factor of three did not have an appreciable effect on these observations although this is
18 presumably a power supply limitation since experiments on both a backward facing step and airfoil leading edge
show reduction in control authority at high energy levels (Akins et al. 2015; Komuro et al. 2018). The reduction
19
in control authority during high amplitude forcing is somewhat counterintuitive and seems to be a characteristic
20

us
of AFC with thermal energy deposition. The cause of this appears related to the competition between thermal
21
perturbations and thermal stabilization (Little 2018). The latter is well-established in mixing layers with heat
22 release (McMurtry et al. 1986; Hermanson et al. 1987; Hermanson and Dimotakis 1989).
23 From a fluid dynamics perspective, the observations presented herein are very much in line with the expected
24 response of the separated flow to actuation. As previously discussed, Wu et al. (1998) predicted the baseline
25 behavior of the NACA 0012 for various angles of attack and Reynolds numbers. The existence of a “low frequency
26 regime” was identified and is commensurate with the experimental findings of Zaman et al. (1989a) who attributed
27
28
29
an
this to the transitional nature of the boundary layer near separation. Given this sensitivity to natural disturbances,
it is expected that AFC can have a strong influence on the flow in this regime and indeed that is the case. Single
pulse forcing behaves similarly in that even though the perturbation is quite short relative to other actuators
30 (Tpulse/Tconv= 2.4x10-8), the flow responds in the same way. This further emphasizes that it is the flow physics that
31 govern the global response to actuation.
32 The novelty of ns-DBDs (and thermal energy deposition in general) for AFC is not in the flow response itself,
dM
33 but rather the ability to generate this response at conditions that are challenging if not impossible for more
34 traditional actuation techniques (e.g. ZNMF momentum addition). The results presented in this work are at the
35 high Reynolds number boundary of the low frequency regime predicted by Wu et al. (1998), but leading edge
36 airfoil separation control authority has been demonstrated at M=0.4 and Re=2,300,000 (Kelley et al. 2014) albeit
37 for a different airfoil. It is expected that similar results can be produced at lower Mach and Reynolds numbers
38 than surveyed here affecting both boundary layer transition and excitation of large coherent structures. It is
39 noteworthy to mention that Kelley et al. (2014) also demonstrated control using ac-DBDs at these conditions.
40 However, demonstrations of this type using ZNMF momentum-based forcing are not widespread especially at
41 higher subsonic Mach numbers.
While the global response to actuation is dictated by the flow physics, the local behavior near the forcing
42
location is very much an open question. The precise mechanism by which ns-DBDs and thermal perturbations
43
pte

initiate flow control is still elusive. The weak compression waves produced by the discharge have been disregarded
44
as a control mechanism as has the modest induced velocity. It appears that a local viscosity or density change is
45 responsible for the control authority (Correale et al., 2014), but further collaborative efforts between experiments,
46 computations and theory are necessary for full clarification. The thickness of the separating boundary/shear layer
47 has a significant influence on the control authority and, at present, quantitative amplitude scaling arguments for
48 ns-DBDs and thermal perturbations in general are not established. However, energy values of ~1 mJ/cm/pulse are
49 widely seen to be sufficient for exciting shear layers having an initial momentum thickness of ~0.1mm. The
50 freestream velocity which is often a limiting factor for momentum-based actuators seems of secondary
51
ce

importance. Therein lies the reason for the vast amount of leading edge airfoil separation control studies using ns-
52 DBDs. The separating boundary layer on the airfoil in stall is quite thin and possibly laminar or transitional making
53 it highly amendable to thermal energy deposition. Nevertheless, this does not preclude ns-DBD utility in turbulent
54 flows provided the separated boundary/shear layer is sufficiently thin (Lehmann et al. 2016).
55 The most important open question related to ns-DBDs and thermal energy deposition for AFC in general
56 concerns the proper amplitude scaling. The dimensional observation posed in the preceding paragraph should be
Ac

57 advanced to allow prediction of the required amplitude for exciting the shear layer instability and controlling a
58 given flow. A more formal understanding of the competition between thermal perturbations and thermal
59 stabilization is required to fully define this scaling argument. The foundation for much of what is known about
60 AFC has been developed using insight from nominally 2D turbulent shear layers. The physics of 3D forcing for

16
Page 17 of 19 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1

1
2
3 AFC are less established and ripe for investigation using ns-DBDs. The high amplitude capability of thermal
4 energy deposition allows for studies of flow physics at conditions that have not yet been adequately explored to
5 date (e.g. Samimy et al. (2018)). Finally, the use of ns-DBDs and AFC in general as enabling tools for the study
6 of other problems such as vortex body interaction (Weingaertner et al. 2018) and the impulse response of complex
7 turbulent flows (Sinha et al. 2012) should not be understated.
8

pt
9 6. Acknowledgements
10
11 This work has been primarily supported by the U.S. Army Research Office Young Investigator Program
12 (W911NF-14-1-0662 monitored by Dr. Matthew Munson), U.S. Army Research Office High School
13 Apprenticeship Program/University Research Apprenticeship Program, Arizona-NASA Space Grant Consortium,

cri
14 Department of Aerospace and Mechanical Engineering and the College of Engineering at the University of
15 Arizona. Timothy Ashcraft was supported by the Army Advanced Civil Schooling Program and the Civil and
16 Mechanical Engineering Department at the United States Military Academy. The authors wish to thank
17 undergraduate students Kenneth Decker, Marcel Dengler, Jorge Castro Maldonado, Zachary Morrett and Zachary
18 Wellington for their assistance.
19
20 References

us
21
22 Akins D, Singh A, Little J (2015) Effects of Pulse Energy on Shear Layer Control using Surface Plasma
23 Discharges. AIAA Paper 2015-3344
24 Albrecht T, Weier T, Gerbeth G, Monnier B, Williams D (2015) Separated Flow Response to Single Pulse
25 Actuation. AIAA Journal 53[1]:190-199
Amitay M, Glezer A (2002) Controlled transients of flow reattachment over stalled airfoils. International
26
27
28
29
an
Journal of Heat and Fluid Flow 23[5]:690-699
Amitay M, Glezer A (2006) Flow Transients induced on a 2D Airfoil by Pulse-modulated Actuation.
Experiments in Fluids 40:329-331
Bayoda K, Benard N, Moreau E, (2015a) Single Nanosecond Dielectric Barrier Discharge and Pulsed Sliding
30 Discharge Applied to Separation Control Along a NACA 0015 Airfoil. AIAA Paper 2015-2344
31 Bayoda K, Benard N, Moreau E, (2015b) Nanosecond pulsed sliding dielectric barrier discharge plasma actuator
32 for airflow control: Electrical, optical, and mechanical characteristics Journal of Applied Physics 118,
dM
33 063301
34 Brzozowski DP, K. Woo GT, Culp JR, Glezer A (2010) Transient Separation Control Using Pulse-Combustion
35 Actuation. AIAA Journal 48[11]:2482-2490 DOI 10.2514/1.45904
36 Clifford C, Singhal A, Samimy M (2016) Flow Control over an Airfoil in Fully Reversed Condition Using
37 Plasma Actuators. AIAA Journal 54[1]
38 Correale G, Michelis T, Ragni D, Kotsonis M, Scarano F, (2014) Nanosecond-pulsed plasma actuation in
39 quiescent air and laminar boundary layer. Journal of Physics D: Applied Physics 47 105201
40 Corke T, Enloe C, Wilkinson S (2010) Dielectric Barrier Discharge Plasma Actuators for Flow Control. Annual
41 Review of Fluid Mechanics 42:505-529
42 Darabi A, Wygnanski I (2004a) Active management of naturally separated flow over a solid surface. Part 2.
43 The separation process. Journal of Fluid Mechanics 510:131-144
pte

44 Darabi A, Wygnanski I (2004b) Active Management of Naturally Separated Flow over a Solid Surface. Part 1.
45 The Forced Reattachment Process. Journal of Fluid Mechanics 510:105-129
46 Dawson R, Little J (2013) Characterization of nanosecond pulse driven dielectric barrier discharge plasma
47 actuators for aerodynamic flow control. Journal of Applied Physics 113[10]:103302-103310
48 DeBlauw B, Elliott G, Dutton C (2014) Active Control of Supersonic Base Flows with Electric Arc Plasma
Actuators. AIAA Journal 52[7]:1502-1517 DOI 10.2514/1.j052585
49
Durasiewicz C, Singh A, Little JC (2018) A Comparative Flow Physics Study of Ns-DBD vs Ac-DBD Plasma
50
Actuators for Transient Separation Control on a NACA 0012 Airfoil. In: 2018 AIAA Aerospace
51
ce

Sciences Meeting: American Institute of Aeronautics and Astronautics

52 Enloe C, McLaughlin T, VanDyken R, Kachner K, Jumper E, Corke T, Post M, Haddad O (2004) Mechanisms
53 and Responses of a Single Dielectric Barrier Plasma Actuator: Geometric Effects. AIAA Journal
54 42[3]:595-604
55 Glezer A, Amitay M, Honohan A (2005) Aspects of Low- and High-Frequency Actuation for Aerodynamic
56 Flow Control. AIAA Journal 43[7]:1501-1511
Ac

57 Greenblatt D, Wygnanski I (2000) The Control of Flow Separation by Periodic Excitation. Progress in
58 Aerospace Sciences 36:487-545
59 Hermanson JC, Dimotakis PE (1989) Effects of Heat Release in a Turbulent, Reacting Shear Layer. Journal of
60 Fluid Mechanics 199:333-375

17
 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1 Page 18 of 19

1
2
3 Hermanson JC, Mungal MG, Dimotakis PE (1987) Heat Release Effects on Shear-Layer Growth and
4 Entrainment. AIAA Journal 25[4]:578-583
5 Jukes T, Choi K, Johnson G, Scott S (2006) Characterization of Surface Plasma-Induced Wall Flows Through
6 Velocity and Temperature Measurements. AIAA Journal 44[4]:764-771
7 Kelley C, Bowles P, Cooney J, He C, Corke T, Osborne B, Silkey J, Zehnle J (2014) Leading-Edge Separation
8 Control Using Alternating-Current and Nanosecond-Pulse Plasma Actuators. AIAA Journal

pt
9 52[9]:1871-1884
10 Komuro A, Takashima K, Suzuki K, Kanno S, Bhandari S, Nonomura T, Kaneko T, Ando A, Asai K (2018)
11 Evaluation of discharge energy and lift recovery of NACA0015 airfoil controlled by nanosecond-pulse-
12 driven plasma actuator. In: 2018 AIAA Aerospace Sciences Meeting: American Institute of
13 Aeronautics and Astronautics

cri
14 Lazar E, Elliot G, Glumac N (2008) Control of the Shear Layer Above a Supersonic Cavity Using Energy
15 Deposition. AIAA Journal 46[12]:2987-2997
16 Lehmann R, Akins D, Little J (2016) Effects of Nanosecond Pulse Driven Plasma Actuators on Turbulent Shear
17 Layers. AIAA Journal 54[2]:637-651
18 Leonov S, Adamovich I, Soloviev V (2016) Dynamics of near-surface electric discharges and mechanisms of
theirinteraction with the airflow. Plasma Sources Science and Technology 25
19
Little J (2018) Localized Thermal Perturbations for Control of Turbulent Shear Flows. AIAA Journal invited
20

us
submission to the Special Issue on Flow Control
21
Little J, Takashima K, Nishihara M, Adamovich I, Samimy M (2012) Separation Control with Nanosecond-
22 Pulse-Driven Dielectric Barrier Discharge Plasma Actuators. AIAA Journal 50[2]:350-365
23 Mathis R, Collin E, Delville J, Bonnet JP (2007) Analysis of a plane turbulent mixing layer manipulated by a
24 localized forced separation. Journal of Turbulence 8[56]:1-14
25 Mathis R, Lebedev A, Collin E, Delville J, Bonnet J (2009) Experimental Study of Transient Forced Turbulent
26
27
28
29
an
Separation and Reattachment on a Bevelled Trailing Edge. Experiments in Fluids 46:131-146
McMurtry P, Jou W, Riley J, Metcalfe R (1986) Direct Numerical Simulations of a Reacting Mixing Layer with
Chemical Heat Release. AIAA Journal 24[6]:962-970
Melton L, Pack, Schaeffler NW, Yao C-S, Seifert A (2005) Active Control of Flow Separation from
30 Supercritical Airfoil Leading-Edge Flap Shoulder. Journal of Aircraft 42[5]:1142-1149
31 Rethmel C, Little J, Takashima K, Sinha A, Adamovich I, Samimy M (2011) Flow Separation Control Using
32 Nanosecond Pulse Driven DBD Plasma Actuators. International Journal of Flow Control 3[4]:213-232
dM
33 Roupassov D, Nikipelov A, Nudnova M, Starikovskii A (2009) Flow Separation Control by Plasma Actuator
34 with Nanosecond Pulsed-Periodic Discharge. AIAA Journal 47[1]:168-185
35 Samimy M, Kim J, Kearney-Fischer M, Sinha A (2010) Acoustic and Flow Fields of an Excited High Reynolds
36 Number Axisymmetric Supersonic Jet. Journal of Fluid Mechanics 656:507-529
37 Samimy M, Webb N, Crawley M (2018) Excitation of Free Shear-Layer Instabilities for High-Speed Flow
38 Control. AIAA Journal 56[5]:1770-1791 DOI 10.2514/1.J056610
39 Seifert A, Darabi A, Wygnanski I (1996) Delay of Airfoil Stall by Periodic Excitation. Journal of Aircraft
40 33[4]:691-698
41 Seifert A, Tilmann C (2009) Fixed Wing Airfoil Applications. In: Fundamentals and Applications of Modern
Flow Control. eds Joslin, R and Miller, D). Vol. 231, pp. 231-257, Reston, VA: AIAA
42
Singhal A, Castañeda D, Webb N, Samimy M (2017) Control of Dynamic Stall over a NACA 0015 Airfoil
43
pte

Using Plasma Actuators. AIAA Journal 56[1]:78-89 DOI 10.2514/1.J056071

44
Sinha A, Alkandry H, Kearney-Fischer M, Samimy M, Colonius T (2012) The impulse response of a high-speed
45 jet forced with localized arc filament plasma actuators. Physics of Fluids 24[12]:125104-125120
46 Starikovskii A, Anikin N, Kosarev I, Mintoussov E, Nudnova M, Rakitin A, Roupassov D, Starikovskaia S,
47 Zhukov V (2008) Nanosecond-Pulsed Discarges for Plasma-Assisted Combustion and Aerodynamics.
48 Journal of Propulsion and Power 24[6]:1182-1197
49 Starikovskii A, Nikipelov A, Nudnova M, Roupassov D (2009) SDBD plasma actuator with nanosecond pulse-
50 periodic discharge. Plasma Science and Technology 18:17
51
ce

Sung Y, Kim W, Mungal M, Cappelli M (2006) Aerodynamic Modification of Flow over Bluff Objects by
52 Plasma Actuation. Experiments in Fluids Vol. 41:479-486
53 Takashima K, Zuzeek Y, Lempert W, Adamovich I (2011) Characterization of a Surface Dielectric Barrier
54 Discharge Plasma Sustained by Repetitive Nanosecond Pulses. Plasma Sources Science and
55 Technology 20[055009]
56 Utkin YG, Keshav S, Kim J-H, Kastner J, Adamovich IV, Samimy M (2007) Development and Use of
Ac

57 Localized Arc Filament Plasma Actuators for High-Speed Flow Control. Journal of Physics D: Applied
58 Physics 40[3]:685-694
59 Webb N, Samimy M (2017) Control of Supersonic Cavity Flow Using Plasma Actuators N.Webb∗. AIAA
60 Journal 55[10]:3346-3355

18
Page 19 of 19 AUTHOR SUBMITTED MANUSCRIPT - PSST-102532.R1

1
2
3 Weingaertner A, Tewes P, Little JC (2018) Parallel Vortex Body Interaction Enabled by Active Flow Control.
4 In: 2018 Flow Control Conference: American Institute of Aeronautics and Astronautics
5 Williams DR, Tadmor G, Colonius T, Kerstens W, Quach V, Buntain S (2009) Lift Response of a Stalled Wing
6 to Pulsatile Disturbances. AIAA Journal 47[12]:3031-3037 DOI 10.2514/1.45407
7 Wu J, Lu X, Denny A, Fan M, Wu J (1998) Post-stall Flow Control on an Airfoil by Local Unsteady Forcing.
8 Journal of Fluid Mechanics 371:21-58

pt
9 Yugulis K, Hansford S, Gregory J, Samimy M (2014) Control of High Subsonic Cavity Flow Using Plasma
10 Actuators. AIAA Journal 52[7]:1542-1554
11 Zaman, McKinzie, Rumsey (1989) A natural low-frequency oscillation of the flow over an airfoil near stalling
12 conditions. Journal of Fluid Mechanics 202:403-442
13

cri
14
15
16
17
18
19
20

us
21
22
23
24
25
26
27
28
29
an
30
31
32
dM
33
34
35
36
37
38
39
40
41
42
43
pte

44
45
46
47
48
49
50
51
ce

52
53
54
55
56
Ac

57
58
59
60

19