Computational and experimental characterization of a liquid jet plunging into a quiescent pool at shallow inclination

A circular water jet (Re = 1.6 × 10⁵; We = 8.8 × 10³) plunging at shallow angles (θ ≈ 12.5°) into a quiescent pool is investigated computationally and experimentally. A surprising finding from the work is that cavities, of the order of jet diameter, are formed periodically in the impact location, even though the impinging flow is smooth and completely devoid of such a periodicity. Computational prediction of these frequencies was compared with experimental findings, yielding excellent agreement. The region in the vicinity of the impact is characterized by strong churning due to splashing and formation of air cavities. Measured velocity profiles indicate a concentration of momentum beneath the free surface slightly beyond the impact location (X/D_j ≈ 14), with a subsequent shift towards the free surface further downstream of this point (X/D_j ≈ 30). This shift is due primarily to the action of buoyancy on the cavity/bubble population. Comparisons of the mean velocity profile between simulations and experiments are performed, yielding good agreement, with the exception of the relatively small churning flow region. Further downstream (X/D_j ≳ 40), the flow develops mostly due to diffusion and the location of peak velocity coincides with the free surface. In this region, the free surface acts as an adiabatic boundary and restricts momentum diffusion, causing the peak velocity to occur at the free surface.     

Study on premixed combustion in cylindrical micro combustors: Transient flame behavior and wall heat flux

The micro combustor is a key component of the micro thermophotovoltaic (TPV) system. Improving the wall temperature of the micro combustor is an effective way to elevate the system efficiency. An experimental study on the wall temperature and radiation heat flux of a series of cylindrical micro combustors (with a backward-facing step) was carried out. For the micro combustors with d = 2 mm, the regime of successful ignition (under the cold wall condition) was identified for different combustor lengths. Acoustic emission was detected for some cases and the emitted sound was recorded and analyzed. Under the steady-state condition, the effects of the combustor diameter (d), combustor length (L), flow velocity (u₀) and fuel–air equivalence ratio (Ф) on the wall temperature distribution were investigated by measuring the detailed wall temperature profiles. In the case that the micro combustor is working as an emitter, the optimum efficiency was found at Ф ≈ 0.8, independent of the combustor dimensions (d and L) and the flow velocity. Under the experimental conditions employed in the present study, the positions of the peak wall temperature were found to be about 8–11 mm and 4–6 mm from the step for the d = 3 mm and d = 2 mm micro combustors, respectively, which are 8–11 and 8–12 times of their respective step heights. This result suggests that the backward-facing step employed in the combustor design is effective in stabilizing the flame position.     

Lattice Boltzmann simulation of the convective heat transfer from a stream-wise oscillating circular cylinder

In this paper, we studied the convective heat transfer from a stream-wise oscillating circular cylinder. Two dimensional numerical simulations are conducted at Re = 100–200, A = 0.1–0.4 and F = f_o/f_s = 0.2–3.0 with the aid of the lattice Boltzmann method. In particular, detailed attentions are paid on the extensive numerical results elucidating the influence of oscillation frequency, oscillation amplitude and Reynolds number on the time-average and RMS value of the Nusselt number. Over the ranges of conditions considered herein, the heat transfer characteristics are observed to be influenced in an intricate manner by the value of the oscillation frequency (F), oscillation amplitude (A) and Reynolds number (Re). Firstly, the heat transfer is enhanced when the cylinder oscillates stream-wise with small amplitude and low frequency, while it will be reduced by large amplitude and high frequency. Secondly, the average Nusselt number (Nu (ave)) decreases against the increasing value of oscillation frequency, while the RMS value of the Nusselt number, Nu (RMS), displays an opposite trend. Third, we obtained a similar frequency effect on the heat transfer over the range of Reynolds numbers investigated in this paper. In addition, detailed analyses on phase portraits, energy spectrum are also made.     

Effects of the jet-to-crossflow momentum ratio on a sonic jet into a supersonic crossflow

Numerical investigation of a transverse sonic jet injected into a supersonic crossflow was carried out using large-eddy simulation for a free-stream Mach number M = 1.6 and a Reynolds number Re = 1.38 × 10⁵ based on the jet diameter. Effects of the jet-to-crossflow momentum ratio on various fundamental mechanisms dictating the intricate flow phenomena, including flow structures, turbulent characters and frequency behaviors, have been studied. The complex flow structures and the relevant flow features are discussed to exhibit the evolution of shock structures, vortical structures and jet shear layers. The strength of the bow shock increases and the sizes of the barrel shock and Mach disk also increase with increasing momentum ratio. Turbulent characters are clarified to be closely related to the flow structures. The jet penetration increases with the increase of the momentum ratio. Moreover, the dominant frequencies of the flow structures are obtained using spectral analysis. The results obtained in this letter provide physical insight in understanding the mechanisms relevant to this complex flow.     

The effect of sidewalls on rectangular jets

An experimental study is presented regarding the influence of sidewalls on the turbulent free jet flow issuing from a smoothly contracting rectangular nozzle of aspect ratio 15. “Sidewalls” are two parallel plates, flush with each of the slots’ short sides, practically establishing bounding walls extending the nozzle sidewalls in the downstream direction. Measurements of the streamwise and lateral velocity mean and turbulent characteristics have been accomplished, with an x-sensor hot wire anemometer, up to an axial distance of 35 nozzle widths, for jets with identical inlet conditions with and without sidewalls. Centreline measurements for both configurations have been collected for three Reynolds numbers, Re_D = 10,000, 20,000 and 30,000. For Re_D = 20,000 measurements in the transverse direction were collected at 13 different downstream locations in the range, x = 0–35 nozzle widths, and in the spanwise direction at three different downstream locations, x = 2, 6 and 25 nozzle widths.Results indicate that, the two jet configurations (with and without sidewalls) produce statistically different flow fields. Sidewalls do not lead to the production of a 2D flow field as undulations in the spanwise mean velocity distribution indicate. They do increase the two-dimensionality of the jet increasing the longevity of 2D spanwise rollers structures formed in the initial stages of entrainment, which are responsible for the convection of longitudinal momentum towards the outer field, establishing larger streamwise mean velocities at the jet edges. In the near field, up to 25 nozzle widths, lower outward lateral velocities in the presence of the sidewalls are held responsible for the decrease of turbulent terms including rms of velocity fluctuations and Reynolds stresses. Skewness factors increase monotonically across the shear layers from negative values to positive forming sharp peaks at the outer edges of the jet, illustrative of the presence of well defined 2D roller structures in the jet with sidewalls.     

Flow instability in baffled channel flow

Flow instability in baffled channel flow, where thin baffles are mounted on both channel walls periodically in the direction of the main flow, has been numerically investigated. The geometry considered here can be regarded as a simple model for finned heat exchangers. The aim of this investigation is to understand how baffle interval (L) and Reynolds number (Re) influence the flow instability. With a fixed baffle length of one quarter of channel height (H), ratios of baffle interval to channel height (R_B = L/H) between 1 and 4 are considered. The critical Reynolds number of the primary instability, a Hopf bifurcation from steady flow to time-periodic flow, turned out to be minimum when R_B = 3.08. The friction factor (f) is strongly correlated with the critical Reynolds number for R_B ⩽ 2.5. For the particular cases of R_B = 1.456 and R_B = 1.0, we performed Floquet stability analysis in order to study the secondary instability through which time-periodic two-dimensional flow bifurcates into three-dimensional flow. The results obtained in this investigation are in good agreement with those computed from full simulations, and shed light on understanding and controlling flow characteristics in a finned heat exchanger, quite beneficial to its design.     

Droplet size and velocity characteristics of water-air impinging jet atomizer

A water-air impinging jets atomizer is investigated in this study, which consists of flow visualization using high speed photography and mean droplet size and velocity distribution measurements of the spray using Phase Doppler Anemometry (PDA). Topological structures and break up details of the generated spray in the far and near fields are presented with and without air jet and for an impinging angle of 90°. Spray angle increases with the water jet velocity, air flow rate and impinging angle. PDA results indicate that droplet size is smallest in the spray center, with minimum value of Sauter mean diameter (SMD) of 50 µm at the air flow rate of Q_m = 13.50 g/min. SMD of droplets increases towards the spray outer region gradually to about 120 µm. The mean droplet velocity component W along the air-jet axis is highest in the spray center and decreases gradually with increasing distance from the spray center. SMD normalized by the air nozzle diameter is found firstly to decrease with gas-to-liquid mass ratio (GLR) and air-to-liquid momentum ratio (ALMR) and then remain almost constant. Its increasing with aerodynamic Weber number indicates an exponential variation. The study sheds light on the performance of water-air impinging jets atomizers providing useful information for future CFD simulation works.     

Experimental study of parallel and inclined turbulent wall jets

An experimental study of the flow field in a two-dimensional wall jet has been conducted. All measurements were carried out using hot-wire anemometry. The experimental facility has a rectangular slot nozzle of high aspect ratio l/b = 100 (where l and b are the length and height slot, respectively). Mean velocities and Reynolds stresses were determined with three nozzle Reynolds numbers (Re = 1 × 10⁴, 2 × 10⁴ and 3 × 10⁴) and four different inclination angles between the wall and the flow velocity at the nozzle (β = 0°, 10°, 20° and 30°). Results indicate that all wall jets are self-preserving in the developed region. Normal to the wall two regions can be identified: one similar to a plane free jet and the other similar to a boundary layer. Downstream the interaction between these two regions creates a mixed or third region. The logarithmic region increases with the distance from the nozzle and with the Reynolds number. For the inclined wall jet, the spreading rate expressed in terms of jet half-width or maximum velocity decay with respect to the streamwise distance, asymptotes to a linear law. The streamwise locations where the jet becomes self-similar are farther from the exit than in parallel wall jet. The slope of both half-width and maximum velocity decay in the developed region are affected by both wall jet inclination angle and nozzle exit Reynolds number.     

Static wall layers in plane channel displacement flows

We study steady and pulsating displacement flows of a Bingham fluid by a Newtonian fluid, along a plane channel. For sufficiently large yield stress a static residual wall layer can result during the displacement. The flow is parameterised by the Reynolds number (Re), the Bingham number (B) and the viscosity ratio (M). Perhaps intuitively, thicker layers are found with larger M and at lower Re. The residual layer is formed on the advective timescale of the displacement but drains on a slower timescale governed by M. For larger M truly stationary layers are only found for large t when the layer has thinned sufficiently to become static. Increased Re results in increased energy production locally around the finger. For large enough Re the energy production can play a significant role in yielding the fluid. As the energy production rate increases it also becomes focused around the corner or shoulder region of the front, and spreads axially along the initial part of the residual layer. This causes fluid to yield increasingly far behind the front and allows for the layer to thin. As B increases the static layer tends to decrease (see also [1], [2]). At small Re the static layer thickness appears to be independent of M. At large Re the layer thickness is dependent on M and decreases asymptotically to a constant value as B → ∞.For pulsating displacement flow rates, Q(t) = 2(1 + Asin  ωt) : A ∈ [0, 1) we study two ranges: ωRe ? 2π and ωRe ? 2π. For the viscous regime (ωRe ? 2π) a pseudo-steady 1D model predicts that the residual layer should remain static for 3(1 + Asin  ωt) < MB. In practice we find that partial mobilisation of the residual layer occurs even when this inequality is satisfied, but not if MB becomes significantly larger than 3(1 + A). For ωRe ? 2π we mobilise the layer for significantly larger values of MB and at smaller A, than in the viscous regime. This effect is traced to the occurrence of out-of-phase velocity fluctuations in the displacing fluid within a wall layer close to the interface.     

Flow evolution of a turbulent submerged two-dimensional rectangular free jet of air. Average Particle Image Velocimetry (PIV) visualizations and measurements

The paper presents average flow visualizations and measurements, obtained with the Particle Image Velocimetry (PIV) technique, of a submerged rectangular free jet of air in the range of Reynolds numbers from Re = 35,300 to Re = 2200, where the Reynolds number is defined according to the hydraulic diameter of a rectangular slot of height H. According to the literature, just after the exit of the jet there is a zone of flow, called zone of flow establishment, containing the region of mixing fluid, at the border with the stagnant fluid, and the potential core, where velocity on the centerline maintains a value almost equal to the exit one. After this zone is present the zone of established flow or fully developed region. The goal of the paper is to show, with average PIV visualizations and measurements, that, before the zone of flow establishment is present a region of flow, never mentioned by the literature and called undisturbed region of flow, with a length, L_U, which decreases with the increase of the Reynolds number. The main characteristics of the undisturbed region is the fact that the velocity profile maintains almost equal to the exit one, and can also be identified by a constant height of the average PIV visualizations, with length, L_CH, or by a constant turbulence on the centerline, with length L_CT. The average PIV velocity and turbulence measurements are compared to those performed with the Hot Film Anemometry (HFA) technique. The average PIV visualizations show that the region of constant height has a length L_CH which increases from L_CH = H at Re = 35,300 to L_CH = 4–5H at Re = 2200. The PIV measurements on the centerline of the jet show that turbulence remains constant at the level of the exit for a length, L_CT, which increases from L_CT = H at Re = 35,300 to L_CT = 4–5H at Re = 2200. The PIV measurements show that velocity remains constant at the exit level for a length, L_U, which increases from L_U = H at Re = 35,300 to L_U = 6H at Re = 2200 and is called undisturbed region of flow. In turbulent flow the length L_U is almost equal to the lengths of the regions of constant height, L_CH, and constant turbulence, L_CT. In laminar flow, Re = 2200, the length of the undisturbed region of flow, L_U, is greater than the lengths of the regions of constant height and turbulence, L_CT = L_CH = 4–5H. The average PIV and HFA velocity measurements confirm that the length of potential core, L_P, increases from L_P = 4–5H at Re = 35,300 to L_P = 7–8H at Re = 2200, and are compared to the previous experimental and theoretical results of the literature in the zone of mixing fluid and in the fully developed region with a good agreement.     

Surface responses induced by point load or uniform traction moving steadily on an anisotropic half-plane

Surface responses induced by point load or uniform traction moving steadily with subsonic speed on an anisotropic half-plane boundary are investigated. It is found that the effects of the material constant on surface displacements are through matrices L^?1(v) and S(v)L^?1(v), while those on surface stress components are through matrices Ω(v) and Γ(v). Explicit expressions for the elements of these four matrices are expressed in terms of elastic stiffness for general anisotropic materials. The special cases of monoclinic materials with symmetry plane at x₁ = 0, x₂ = 0 and x₃ = 0, and the case for orthotropic materials are all deduced. Results for isotropic material may be recovered from present results. For monoclinic materials with a plane of symmetry at x₃ = 0, two of the elements of matrix Ω(v) are found to be independent of subsonic speed.     

Turbulent boundary layers over sparsely-spaced rod-roughened walls

Direct numerical simulations (DNSs) of spatially developing turbulent boundary layers (TBLs) over sparsely-spaced two-dimensional (2D) rod-roughened walls were performed. The rod elements were periodically arranged along the streamwise direction with pitches of p_x/k = 8, 16, 32, 64 and 128, where p_x is the streamwise spacing of the rods, and k is the roughness height. The Reynolds number based on the momentum thickness was varied from Re_θ = 300–1400, and the height of the roughness element was k = 1.5θ_in, where θ_in is the momentum thickness at the inlet. The characteristics of the TBLs, such as the friction velocity, mean velocity, and Reynolds stresses over the rod-roughened walls, were examined by varying the spacing of the roughness features (8 ⩽ p_x/k ⩽ 128). The outer-layer similarity between the rough and smooth walls was established for the sparsely-distributed rough walls (p_x/k ⩾ 32) based on the profiles of the Reynolds stresses, whereas those are not for p_x/k = 8 and 16. Inspection of the interaction between outer-layer large-scale motions and near-wall small-scale motions using two-point amplitude modulation (AM) covariance showed that modulation effect of large-scale motions on near-wall small-scale motions was strongly disturbed over the rough wall for p_x/k = 8 and 16. For p_x/k ⩾ 32, the flow that passed through the upstream roughness element transitioned to a smooth wall flow between the consecutive rods. The strong influence of the surface roughness in the outer layer for p_x/k = 8 and 16 was attributed to large-scale erupting motions by the surface roughness, creating both upward shift of the near-wall turbulent energy and active energy production in the outer layer with little influence on the near-wall region.     

Turbulent channel flow of dilute polymeric solutions: Drag reduction scaling and an eddy viscosity model

Direct numerical simulation of viscoelastic turbulent channel flows up to the maximum drag reduction (MDR) limit has been performed. The simulation results in turn have been used to develop relationships between the flow and fluid rheological parameters, i.e. maximum chain extensibility, Reynolds number, Re_τ, and Weissenberg number, We_τ and percent drag reduction (%DR) as well as the slope increment of the mean velocity profile. Moreover, based on the trends observed in the mean velocity profile and the overall momentum balance three different regimes of drag reduction (DR), namely, low drag reduction (LDR; 0 ≤ %DR ≤ 20), high drag reduction (HDR; 20 ≤ %DR ≤ 52) and MDR (52 ≤ %DR ≤ 74) have been identified and mathematical expressions for the eddy viscosity in these regimes are presented. It is found that both in LDR and HDR regimes the eddy viscosity varies with the distance from the channel wall. However, in the MDR regime the ratio of the eddy viscosity to the Newtonian one tends to a very small value around 0.1 within the channel. Based on these expressions a procedure that relies on the DNS predictions of the budgets of momentum and viscoelastic shear stress is developed for evaluating the mean velocity profile.     

Some exact solutions for rotating flows of a generalized Burgers’ fluid in cylindrical domains

The velocity field and the adequate shear stress corresponding to the flow of a generalized Burgers’ fluid model, between two infinite co-axial cylinders, are determined by means of Laplace and finite Hankel transforms. The motion is due to the inner cylinder that applies a time dependent torsional shear to the fluid. The solutions that have been obtained, presented in series form in terms of usual Bessel functions J₁( ? ), J₂( ? ), Y₁( ? ) and Y₂( ? ), satisfy all imposed initial and boundary conditions. Moreover, the corresponding solutions for Burgers’, Oldroyd-B, Maxwell, second grade, Newtonian fluids and large-time transient solutions for generalized Burgers’ fluid are also obtained as special cases of the present general solutions. The effect of various parameters on large-time and transient solutions of generalized Burgers’ fluid is also discussed. Furthermore, for small values of the material parameters, λ₂ and λ₄ or λ₁, λ₂, λ₃ and λ₄, the general solutions corresponding to generalized Burgers’ fluids are going to those for Oldroyd-B and Newtonian fluids, respectively. Finally, the influence of the pertinent parameters on the fluid motion, as well as a comparison between models, is shown by graphical illustrations.     

Wavelet decomposition method decoupled boiling/evaporation oscillation mechanisms over two to three timescales: A study for a microchannel with pin fin structure

Boiling/evaporation heat transfer in a microchannel with pin fin structure was performed with water as the working fluid. Simultaneous measurements of various parameters were performed. The chip wall temperatures were measured by a high spatial-time resolution IR image system, having a sensitivity of 0.02 °C. The flow pattern variations synchronously changed wall temperatures due to ultra-small Bi number. The wavelet decomposition method successfully identified the noise signal and decoupled various temperature oscillations with different amplitudes and frequencies. Three types of temperature oscillations were identified according to heat flux q and mass flux G. The first type of oscillation occurred at q/G < 0.62 kJ/kg. The approximation coefficient of wavelet decomposition decided the dominant cycle period which was ∼3 times of the fluid residence time in the microchannel, behaving the density wave oscillation characteristic. The detail coefficients of wavelet decomposition decided the dominant cycle period, which matched the flow pattern transition determined value well, representing the flow pattern transition induced oscillation. For the second type of oscillation, the wavelet decomposition decoupled the three oscillation mechanisms. The pressure drop oscillation caused the temperature oscillation amplitudes of 5–10 °C and cycle periods of 10–15 s. The density wave oscillation and flow pattern transition induced oscillation are embedded with both the pressure rise and decrease stages of the pressure drop oscillation. The third type of oscillation happened at q/G > 1.13 kJ/kg, having the density wave oscillation coupled with the varied liquid film evaporation induced oscillation. The liquid island, retention bubble induced nucleation sites and cone-shape two-phase developing region are unique features of microchannel boiling with pin fin structure. This study illustrated that pressure drop oscillation and density wave oscillation, usually happened in large size channels, also take place in microchannels. The flow pattern transition and varied liquid film evaporation induced oscillations are specific to microchannel boiling/evaporation flow.     

Dynamic mode decomposition of turbulent cavity flows for self-sustained oscillations

Self-sustained oscillations in a cavity arise due to the unsteady separation of boundary layers at the leading edge. The dynamic mode decomposition method was employed to analyze the self-sustained oscillations. Two cavity flow data sets, with or without self-sustained oscillations and possessing thin or thick incoming boundary layers (Re_D = 12,000 and 3000), were analyzed. The ratios between the cavity depth and the momentum thickness (D/θ) were 40 and 4.5, respectively, and the cavity aspect ratio was L/D = 2. The dynamic modes extracted from the thick boundary layer indicated that the upcoming boundary layer structures and the shear layer structures along the cavity lip line coexisted with coincident frequency space but with different wavenumber space, whereas structures with a thin boundary layer showed complete coherence among the modes to produce self-sustained oscillations. This result suggests that the hydrodynamic resonances that gave rise to the self-sustained oscillations occurred if the upcoming boundary layer structures and the shear layer structures coincided, not only in frequencies, but also in wavenumbers. The influences of the cavity dimensions and incoming momentum thickness on the self-sustained oscillations were examined.     

Evolution of planetary boundary layer under different weather conditions,and its impact on aerosol concentrations

A field experiment was conducted in Tianjin, China from September 9–30, 2010, focused on the evolution of Planetary Boundary Layer (PBL) and its impact on surface air pollutants. The experiment used three remote sensing instruments, wind profile radar (WPR), microwave radiometer (MWR) and micro-pulse lidar (MPL), to detect the vertical profiles of winds, temperature, and aerosol backscattering coefficient and to measure the vertical profiles of surface pollutants (aerosol, CO, SO₂, NO_x), and also collected sonic anemometers data from a 255-m meteorological tower. Based on these measurements, the evolution of the PBL was estimated. The averaged PBL height was about 1000–1300 m during noon/afternoon-time, and 200–300 m during night-time. The PBL height and the aerosol concentrations were anti-correlated during clear and haze conditions. The averaged maximum PBL heights were 1.08 and 1.70 km while the averaged aerosol concentrations were 52 and 17 μg/m³ under haze and clear sky conditions, respectively. The influence of aerosols and clouds on solar radiation was observed based on sonic anemometers data collected from the 255-m meteorological tower. The heat flux was found significantly decreased by haze (heavy pollution) or cloud, which tended to depress the development of PBL, while the repressed structure of PBL further weakened the diffusion of pollutants, leading to heavy pollution. This possible positive feedback cycle (more aerosols → lower PBL height → more aerosols) would induce an acceleration process for heavy ground pollution in megacities.     

Dynamic mode decomposition of turbulent cavity flows for self-sustained oscillations

Self-sustained oscillations in a cavity arise due to the unsteady separation of boundary layers at the leading edge. The dynamic mode decomposition method was employed to analyze the self-sustained oscillations. Two cavity flow data sets, with or without self-sustained oscillations and possessing thin or thick incoming boundary layers (Re_D = 12,000 and 3000), were analyzed. The ratios between the cavity depth and the momentum thickness (D/θ) were 40 and 4.5, respectively, and the cavity aspect ratio was L/D = 2. The dynamic modes extracted from the thick boundary layer indicated that the upcoming boundary layer structures and the shear layer structures along the cavity lip line coexisted with coincident frequency space but with different wavenumber space, whereas structures with a thin boundary layer showed complete coherence among the modes to produce self-sustained oscillations. This result suggests that the hydrodynamic resonances that gave rise to the self-sustained oscillations occurred if the upcoming boundary layer structures and the shear layer structures coincided, not only in frequencies, but also in wavenumbers. The influences of the cavity dimensions and incoming momentum thickness on the self-sustained oscillations were examined.     

Flow boiling heat transfer of R134a,R236fa and R245fa in a horizontal 1.030 mm circular channel

This research focuses on acquiring accurate flow boiling heat transfer data and flow pattern visualization for three refrigerants, R134a, R236fa and R245fa in a 1.030 mm channel. We investigate trends in the data, and their possible mechanisms, for mass fluxes from 200 to 1600 kg/m²s, heat fluxes from 2.3 kW/m² to 250 kW/m² at T_sat = 31 °C and ΔT_sub from 2 to 9 K. The local saturated flow boiling heat transfer coefficients display a heat flux and a mass flux dependency but no residual subcooling influence. The changes in heat transfer trends correspond well with flow regime transitions. These were segregated into the isolated bubble (IB) regime, the coalescing bubble (CB) regime, and the annular (A) regime for the three fluids. The importance of nucleate boiling and forced convection in these small channels is still relatively unclear and requires further research.