· · · Institutional Repository

A shallow water mixing layer is a flow pattern which develops between two adjacent streams with a different velocity. The horizontal dimensions of a shallow water mixing layer are much larger than the depth. This results in a turbulent process in the mixing layer which is characterized by the presumable presence of two distinct turbulent length scales. Small scale turbulence is present due to the bottom friction and large scale structures develop horizontally over the whole vertical interface between the two flows. It is assumed that no intermediate length scales can develop due to the limited water depth. After a certain distance downstream, the large eddies appear not to develop any more. This can be attributed to an equilibrium between the energy supply to the large scale eddies, due to the velocity difference across the mixing layer, and a direct energy loss to bottom turbulence. First, a one-dimensional model is developed to describe the development of flow parameters of the shallow water mixing layer, such as the velocities outside the mixing layer, the water depth and the displacement of the mixing layer. A model matrix is derived on the basis of the equation of continuity and the Navier Stokes equations describing two adjacent flows in a wall-limited shallow water flume. The complex flow pattern in the mixing layer is modelled using an approximation for the mixing layer development and assuming an error function velocity profile over the mixing layer width. The longitudinal pressure gradient, caused by the bottom friction, results in a displacement of the mixing layer to the low velocity side, a smaller velocity difference across the mixing layer and a larger slope downstream. Results based on the error function profile are compared to those of a linear velocity profile. Comparison of the error function velocity profile results with propeller measurements and LDA-measurements shows a reasonable correspondence between model results and measurements. The small differences that occur can possibly be attributed to the neglect of a slope in lateral direction in the model. Second, the turbulent process in a shallow water mixing layer is investigated to reveal the possible presence of two distinct turbulent length scales in the mixing layer. The two distinct length scales are determined by means of measuring the spatial correlations in lateral direction of the mixing layer. Since the spatial correlations measured are caused by small and large scale turbulence, it is assumed that the correlation consists of two parts which are independent from each other. One part is assumed to be caused by the small scale turbulence and the other by the large scale eddies. After separating the two parts, the length scales are determined by integrating the two contributions to the correlation function. Measurements were taken at two elevations at 16 meter downstream from the point where the two flows come together. This results in an order of magnitude difference between the two calculated length scales. At a smaller elevation, the small turbulence length scale is significantly smaller than at a larger elevation. The large turbulence length scale is smaller as well, but this decline is relatively less than the small scale turbulence decline. The decline of the large scale turbulence can possibly be attributed to the length scale determination method used.

Results of sediment transport calculations are often necessary in solving practical coastal engineering problems. (Sediment transport due to waves and currents). Many transport formulae have been proposed in literature in the past. Selection of the proper one while solving a particular problem, is a difficult task for a coastal engineer. In considering sediment transport under wave-current conditions it is worthwhile to make a distinction between two situations, viz.: The fluctuations in the orbital motion have to be fully taken into account in order to find the resulting sediment transport (intra-wave type of description; often: cross-shore sediment transport); - It is sufficient to take time-averaged effects of the waves into account in order to find the resulting sediment transport rate (intra-wave type of description is not required; often: longshore sediment transport). For the longshore sediment transport mode, transport formulae based on time-averaged velocity distributions and time-averaged sediment concentration distributions over the water depth can often be used. The present paper is restricted to this type of formula.

The use of pile groynes to reduce sediment exchange between river and harbour Introduction To reduce siltation in harbours located in rivers or waterways in order to decrease the high costs of dredging and maintenance of its basins, a need of searching measures to decrease the sediment exchange between river and harbour arises. Research One of the measures to reduce the sediment exchange is the use of pile groynes structures located in the river stream, upstream the entrance of the harbour basin. Due to the blocking of the groynes, the velocities of the river will increase. Because of this, the navigability along the river stream can be altered. The objective of this research is to determine the effect of the groyne at the interface harbour-river and the mixing layer which develops along the entrance. Experiments To analyze the effect of the groyne on the mixing layer dynamics, a schematized physical model of a harbour basin was built in a laboratory flume. Different configurations of the pile groyne were performed to devise the effect of different geometries of the groyne. In order to relate the exchange process to velocity distributions, Particle Tracking Velocimetry measurements were performed to obtain the velocity field at the water surface. The analysis of data was approached on the exchange which takes place via the mixing layer and the primary gyre, that are formed at the river-harbour interface and the harbour basin, respectively. On the basis of the experimental results can be concluded that the best configurations to reduce the entrainment of water and the velocity gradient along the entrance are the groynes located close to the upstream corner and with spacing between piles lower than its width. Numerical modelling Several configurations of the groyne were simulated using state of the art of a 2D numerical model, FinLab, in order to perform a comparison between experiments and simulations, providing an indication as to what extent the exchange processes are properly incorporated into this model. A good estimation can be obtained for the case without groyne in relation with the basic structures and turbulence properties of the mixing layer at the interface river-harbour. However, it was not possible to simulate the effect of the groyne by increasing the roughness coefficient at the groyne location. For this reason, some recommendations for further studies about that topic are given.

The general flow pattern of an open channel flow, downstream of a width restriction by two artificial dams, is analysed. A physical Froude-scaled model, under hydraulic rough conditions, with a significant large Reynolds number is used to ensure turbulent flow. Upstream of the dams the flow is uniform in transverse direction, in between and downstream of the narrow part a jet is formed. On both sides of the jet large eddies are formed bounded by the wall, the jet and the dams. Due to the large velocity gradient in transverse direction a mixing layer develops at both sides of the jet. The width of the mixing layer, as expected, grows with the downstream distance and exceeds the water depth. 2D structures are clearly visible by injecting dye. In the mixing layer besides the macro time and spatial scales, the small Taylor and Kolmogorov scales are present. Whereas the macro scales are well represented in the measured data, the small scales are impossible to mark due to limitations of the Doppler device. When there is initial no net momentum in transverse direction present the jet is expected to appear symmetrical. However the jet is aligned to one of the sides every time the model starts to run. The preference for one or the other side seems to be random and cannot be related to momentum in transverse direction in between the dams. During measurements the position of the jet is stationary. The fixed position of the jet during measurements can be related to the Coandă effect. When the flow is disturbed and transverse momentum is added to the upstream flow, the jet can be deflected. The position of the jet and the evolving mixing layers can be related very well to the measured velocities upstream. Due to the limitations of the used momentum balance equation and use of the mean velocity in the bottom friction calculation the measured head loss is large compared to the calculated dissipative terms (bottom friction and Carnot loss).

In onderhavige studie is de meettechniek "Particle Tracking Velocimetry" (PTV) toegepast ten behoeve van het onderzoek naar ondiepe waterstromingen. Het doel van het onderzoek was de meettechniek beschikbaar te krijgen voor studies aan ondiep-waterstromingen en de geschiktheid van de methode voor het meten van grootschalige structuren aan te tonen. Eerst is er een literatuuronderzoek naar de PTV-methode geweest. De methode meet verplaatsingen van deeltjes in een medium over een bepaalde periode. Hiermee kunnen watersnelheden worden bepaald. Vervolgens is de PTV-methode gebruiksklaar gemaakt voor de metingen. Na het opzetten van de PTV-methode is een literatuuronderzoek gedaan naar de werve1s in een menglaag. Hieruit is gekomen dat wervels zijn aan te tonen door samenhang in fluctuaties van snelheden in ruimte en tijd. Daama is er een post processing methode opgezet om de PTV metingen om te zetten naar fluctuaties ten opzichte van een gemiddelde. Om deze methode uit te voeren is er een post processing programma geschreven. Als laatste is bij de opzet van de methode een uitgebreide foutenanalyse geformuleerd. Hiermee is aangetoond dat aanwezige fluctuaties gemeten kunnen worden. Er zijn in twee opstellingen metingen met behulp van de PTV-methode verricht. Eerst is er in een havenmodel gemeten en vervolgens in een ondiep-watermenglaag. De opstelling in het havenmodel is gebruikt om de methode op te zetten en bekend te raken met de methode. De opstelling met de ondiep-watermenglaag is gebruikt als test voor het meten van de structuren. Hierin zijn eveneens LDA metingen verricht waarmee de PTV-methode is vergeleken. Uit de resultaten van de metingen is gebleken dat: het opzetten van de meetmethode is gelukt; het meten van turbulente structuren in de menglaag niet is gelukt. Dit werd veroorzaakt doordat de afstand tussen de deeltjes groter is dan de grootte van de samenhang in de fluctuaties. De belangrijkste oorzaak hiervan was te ondiep water. Hierdoar konden de structuren zich niet in voldoende mate ontwikkelen om deze te kunnen waamemen met deze methode. Aanbevelingen voar de meting aan wervels: voer de experimenten uit met deeltjes die kleiner zijn ell dichter bij elkaar liggen, zodat de wervels meetbaar zijn. Aanbevelingen voor de metingen: er zijn interessante effecten in de metingen zichtbaar geweest, deze kunnen met deze methode goed worden onderzocht.

Summary Of the thesis :’ Mixing and In-situ product removal in micro bioreactors’ by Xiaonan Li The work presented in this thesis is a part of a large cluster project, which was formed between DSM, Organon, Applikon and two university groups (TU Delft and University of Twente), under the ACTS and IBOS program. The aim of this cluster project was to develop a system consisting of parallel bioreactors of 30 to 200 microliter working volume for the cultivation of micro-organisms under well controlled industrially relevant condition (T, pH, DO etc.), and operated as fed-batch reactor in long term (>200h). This platform has the potential to be used for high throughput screening applications for gene identification or the related small scale protein fermentation to increase the protein production process development rate and to reduce the research cost. The development of the platform starts with the design of a single micro-reactor; the single micro-reactor is the integration of well developed sensing system, control system, mixing system and other accessories, like, pumps, valves, adaptors, vessels etc. However many components are not available or not suitable for our application. In this thesis several novel mixing methods, which can provide sufficient mixing in a micro-reactor to satisfy the need of micro-organism fermentation, are developed. Furthermore, microfluidic components are important to facilitate substrate feeding as well as by product removing. An ISPR concept was experimentally demonstrated in this thesis to distinguish a wider scope of micro-reactor applications. One of the main reasons to apply micro systems technology is that compared with traditional reaction (fermentation) technologies, a superior, rapid and sufficient mixing can easily be achieved using micro technologies, especially for those microfluidic devices, which integrated with passive micro structures, with working volume tens of nanoliters. However, the mixing in the microreactor, which with working volme around hundreds microliter, is still be considered as a bottleneck for the high biomass concentration fermentation. In chapter 2, recycle flow mixing (RFM) method was presented. By continuously moving liquid solution from high oxygen concentration area to low oxygen concentration area via multiple fluxes, the system obtains maximum oxygen transfer, which is considered as the bottleneck for high cell density fermentation. Meanwhile, the recycled flows create vigorous convectionin the micro-reactor and obtain good mixing. The mixing performance was experimentally verified with a prototype reactor (with working volume 30 microliter). Under a small recycle flow rate (20 microliter/min), the measured well mixing time was around 800s. However, after taken into account the influence of the recycle tubes (50 microliter), the mixing in the microreactor was considered as comparable as an ideal mixed reactor. The impact of various oxygen transfer abilities on high cell density fermentation was estimated by 2D / 3D CFD simulations. With recycle flow rate 0.001 m/s, the kla value of the microreactor was around 0.023 s-1, which was in the same order of magnitude as a regular stirred tank. This oxygen transfer was sufficient for a high biomass concentration fermentation (Max. biomass concentration > 20 g/l). The mixing performance of RFM method is dependent on the value of the recycle fluxes, therefore, a strong internal micro-pump plays an essential role in the system. To avoid the dependence on the micro-pump development, an alternative micro-mixing method was presented in chapter 3. Oscillation flows, which are created by a central actuator, induce vigorous convection in the micro-reactor(s) to obtain good mixing. The mixing performance within a single reactor was estimated by CFD calculations, a simplified micro-mixing correlation and validated experimentally. With an oscillation frequency (f) of 8.33 Hz, oscillation flow rate (fv) 1000 microliter/min, the experimental well mixing time was 45s; the CFD simulated well mixing time was 37 s; the model calculated well mixing time was 35s. With a stronger oscillation (f=8.33Hz, fv=3000 microliter/min) the well mixing time dropped to 4s.(CFD simulated result & model correlative result) The oscillation mixing method has the potential to be easily integrated with parallel reactorsrrelative This concept has been proven experimentally using a 96-wells micro titer plate and one oscillation pump (f=2.22 Hz, fv=2000 microliter/min). Three parallel reactors followed the same trend and reached to well mixing time at 120s, 122s and 128s, respectively. The comparison of dye distribution results between various tubes indicated a similar mixing behavior in different reactors. Hence, the result show the possibility of using one central actuator to create oscillation fluids to achieve mixing on multi-reactors. Additional experiments have been done with oscillation mixing method to test the influence of the mixing methods on cells viability and influence of the oscillation mixing method on cells suspension. The experiment clearly indicated that compared to the magnetic stirrer mixing method, oscillation mixing method showed less damage on the cells during cell viability test. Homogeneous cell suspension was maintained in the micro bioreactor during overnight oscillation mixing. The characteristics of microfluidic channels for mass transfer were explored in chapter 4. When two liquid streams join into one microchannel with diameter around 150 mm, both streams will behave as laminar flows and run parallel to each other with a stable interface in between. If for certain components there are concentration differences between two streams, over the interface, components can transfer from one stream to another via diffusion. In this chapter the quantitative transfer of glucose between two cocurrent streams was estimated by CFD and experimentally verified. A microchannel has a large surface to volume ratio; therefore, within a short time significant amount of glucose can be transferred from one stream to another. The transfer rate of glucose was measured to be 2.4 - 11.9 nmol /min at a residence time of 54 - 857ms and glucose concentration in the feed stream of a modest 10.4mM. If this transfer would be applied for a fed-batch cultivations in a 100ml microbioreactor, glucose feed rates ranging from 0.26 to 1.3 g/Lreactor/h could be achieved, which is sufficient to perform industrial fermentation processes of fed-batch cultivations at high biomass concentrations. This microfluidic channel also could be used for by-product removal application. The reason by-product needs to be removed is because of the potential risk of the product inhibition, which may cause a decrease in microorganism activity. An implementaed In Situ Product Removal (ISPR) method can circumvent this risk by keeping the dissolved product concentration low in the reactor. Chapter 5 focuses on demonstrating the feasibility of applying a suitable ISPR method on micro-scale bioreactor. Lactic acid was selected as the target chemical. Extraction was selected as the separation method. By pushing a selected extractant (trioctylamine / decanol / dodecane) through a hydrophobic micro hollow fiber, lactic acid is extracted from the aqueous phase into organic phase, and then removed from the microreactor. The micro hollow fiber has the sole task to be the barrier to isolate microorganism from organic phase. The extraction ability was estimated by a model and then validated experimentally. The high specific interfacial area in the micro ISPR system (13.3E+3 m2/m3) shows the advantage of the microextraction for ISPR processes. High ISPR removal rate (1.92e-6 mol/l/s) was obtained experimentally. This removal rate was in the same order of magnitude as the reported lactic acid production rates in mammalian cell cultures (7.09e-7 to 3.7 e-5 mol/l/s). In conclusion, this thesis presents the development of novel micro-mixing methods and the preliminary application of possible In Situ Production Removal (ISPR) methods, leading to the increased applicability of (fed-) batch micro bioreactor for long-term high-biomass concentration fermentation. However, a combined microbioreactor (including sensor, ISPR design and novel mixing design) has yet not been tested experimentally. A number of present challenges is discussed in Chapter 6.

Improved and reproducible heterodyne mixing (noise temperatures of 950 K at 2.5 THz) has been realized with NbN based hot-electron superconducting devices with low contact resistances. A distributed temperature numerical model of the NbN bridge, based on a local electron and a phonon temperature, has been used to understand the physical conditions during the mixing process. We find that the mixing is predominantly due to the exponential rise of the local resistivity as a function of electron temperature.

Tidal straining, which can mathematically be described as the covariance between eddy viscosity and vertical shear of the along-channel velocity component, has been acknowledged as one of the major drivers for estuarine circulation in channelized tidally energetic estuaries. In this paper, the authors investigate the role of lateral circulation for generating this covariance. Five numerical experiments are carried out, starting with a reference scenario including the full physics and four scenarios in which specific key physical processes are neglected. These processes are longitudinal internal pressure gradient forcing, lateral internal pressure gradient forcing, lateral advection, and the neglect of temporal variation of eddy viscosity. The results for the viscosity–shear covariance are correlated across different experiments to quantify the change due to neglect of these key processes. It is found that the lateral advection of vertical shear of the along-channel velocity component and its interaction with the tidally asymmetric eddy viscosity (which is also modified by the lateral circulation) is the major driving force for estuarine circulation in well-mixed tidal estuaries.

Little research was done in the past concerning the propagation of three dimensional effect in shallow wake flow caused by a roughness patch. Today’s research on related subjects is dominated by emerging obstructions in shallow water where the flow can be assumed as (quasi) two dimensional. However the relevance of a submerged obstruction with increased roughness can be found in wake control, oyster reefs, river- and estuary bottoms and heterogeneous land occupancy. To get a better understanding of the consequences of the three dimensionality of the flow structures, experiments are performed in a wide shallow flume to examine these structures. The main objective is to examine whether the wake structure of a roughness patch can be treated as (quasi)-two-dimensional. The objective has been answered by a combination of a literature study and an experiment performed at the faculty’s laboratory. The results show four dominant mechanisms in the wake of a roughness patch: transverse mass flux, bottom friction, mixing layer and the secondary circulation. Based on a momentum balance the transverse mass flux and the bottom friction are the largest contributions to this balance. Although the contribution of the mixing layer and the secondary circulation to the recovery of the wake are of the order of 10%, their influence on the flow structure is more pronounced. The mixing layer is shifted towards the wake centerline due to the presence of a transverse mass flux forming a misalignment between the maximum spanwise Reynolds stress and the position of the wake half width. Since this shift is of limited influence on the position of the secondary circulation, a misalignment if formed between the maximum momentum transport by the secondary circulation and the mixing layer causing a lower streamwise velocity at the edge of wake with respect to the wake of an emerging obstruction. The secondary circulation is responsible for the transport of low momentum fluid towards the edge of the wake near the bottom, and high momentum fluid towards the wake centerline near the surface. This behavior is responsible for the cross gradient in the streamwise velocity profiles as shown by the data obtained. For modeling purposes of well mixed quantities, a (quasi)-two-dimensional approach only holds if the weaker streamwise velocity near the edge of the wake is taken into account. In the case a prediction of depth varying quantities is desired, the cross gradient caused by the secondary circulation needs to be implemented as well which results in the need of a three-dimensional modeling approach.

In stably stratified flows vertical movement of eddies is limited by the fact that kinetic energy is converted into potential energy, leading to a buoyancy displacement scale z B . Our new mixing-length concept for turbulent transport in the stable boundary layer follows a rigid-wall analogy, in the sense that we assume that the buoyancy length scale is similar to neutral length scaling. This implies that the buoyancy length scale is: ℓ B = κ B z B , with κ B ≈ κ, the von Karman constant. With this concept it is shown that the physical relevance of the local scaling parameter z/Λ naturally appears, and that the α coefficient of the log-linear similarity functions is equal to c/κ 2, where c is a constant close to unity. The predicted value α ≈ 1/κ 2 = 6.25 lies within the range found in observational studies. Finally, it is shown that the traditionally used inverse linear interpolation between the mixing length in the neutral and buoyancy limits is inconsistent with the classical log-linear stability functions. As an alternative, a log-linear consistent interpolation method is proposed.

An analysis of field measurements recorded over a tidal cycle in the Rotterdam Waterway is presented. These measurements are the first to elucidate the processes influencing the along-channel current structure and the excursion of the salt wedge in this estuary. The salt wedge structure remained stable throughout the measuring period. The velocity measurements indicate decoupling effects between the layers and that bed-generated turbulence is confined below the pycnocline. The barotropic M4 overtide structure is imposed at the mouth of the estuary, and the generation of M4 overtides within the estuary is found to be relatively small. Internal tidal asymmetry does not make a significant contribution to the M4 velocity frequency band. Instead, the combination of barotropic and baroclinic forcing, in conjunction with the suppression of turbulence at the interface, provides the main explanation for the time dependence and mean structure of the flow in the Rotterdam Waterway. This gives rise to the observed differences in the length of the flood and ebb, in the magnitudes of the flood and ebb velocities, in the length of the slack water periods, and in the timing of the onset of slack water at the surface and near the bed. It results in the formation of distinct exchange flow profiles at the head of the salt wedge around slack water and the creation of maximal velocities at the pycnocline during flood. Advection governs the displacement and structure of the salt wedge since turbulent mixing is suppressed. The tidal displacement of the salt wedge controls the height of the pycnocline above the bed at a particular site. Hence, it controls the height to which bed-generated turbulence can protrude into the water column. Consequently, the authors find asymmetries in the structure of the internal flow, turbulent mixing, and bed stresses that are not related to classical internal tidal asymmetry.

The steam cracking process is an important asset in the hydrocarbon processing industry. The main products are lower olefins and hydrogen, with ethylene being the world's largest volume organic chemical at a worldwide capacity of ~ 120 million tonnes per year. Feed stocks are hydrocarbons such as: ethane, LPG, naphtha's, gas condensates and gas oil. The research goal of this thesis is to search for the intrinsic optimal steam cracking reaction conditions, pushing the olefin yields to the maximum that the fundamental reaction kinetic models allow for. To get to that goal we have: firstly, identified and assessed alternative process concepts published in the literature. Secondly, developed the concepts and software for an equation based modelling tool suitable for optimisation of large scale reaction kinetic models. Thirdly, developed a new reactor concept, d-RMix for homogeneous reactions with distributed feed allocation, product removal and macro-mixing. Fourthly, applied the optimisation tool to the new reactor concept model and an advanced reaction kinetic model for steam cracking, SPYRO(r). For four different feed stocks optimisations of ethylene yield and of ethylene plus propylene yields have been carried out. For the cracking of ethane a linear-concave unconstrained temperature profile with a (free) maximum temperature of ~1260 K proves optimal. For propane and heavier feed stocks an isothermal profile at the upper temperature bound is optimal, with dips at the beginning and the middle of the reaction volume coordinate. For these heavier hydrocarbon feeds a distribution along the reactor volume coordinate does result in higher yields. Having established that the steam cracking chemistry offers a potential for significantly higher olefins yields, these equipment engineering considerations pose a significant challenge to actually realise this potential and arrive at a next generation steam cracking reactor.

The Blum-Hanson property was identified in the 1960's as a property closely related to that of strong mixing. This property is interesting in the field of ergodic theory. It can be shown that if the orbits of all points have the Blum-Hanson property, then the operator is strongly mixing. The converse is not always true and much research has been devoted to determining when the Blum-Hanson property and the strong mixing property are in fact equivalent. After a brief introduction to ergodic theory, Chapter 2 covers all the background and definitions needed to understand the later chapters. Chapter 3 covers the major positive results to date regarding the Blum-Hanson property. Chapter 4 introduces a new class of operators that are interesting for the results in the final chapter, the Generalised Shift Operators. Results for this new class of operators are then used to analyse a number of existing examples in Chapter 5. Here existing results are extended to cover more spaces and a new counter-example is introduced based on a suggestion by László Zsidó.

This study demonstrates the importance of a sophisticated sub-grid model when performing a depth-averaged unsteady RANS simulation of a shallow flow. The reduction of resolution and the spatial dimensions exclude important physical processes as present in three-dimensional turbulence. Especially the effect of the bottom turbulence on the formation of horizontal eddies appears of key importance. A method is proposed to incorporate these effects by means of a kinematic simulation that mimics the residual turbulent fluctuations in a straight channel flow after depth-averaging. This method is developed in the context of the evolution of large eddies in a shallow mixing layer. A comparison with experiments shows that the proposed method works satisfactory. Naturally, it does not fully account for the omission of all 3D-effects.

The vertical rod/tube bundle geometry has a wide variety of industrial applications. Typical examples are the core of light water nuclear reactors (LWR) and vertical tube steam generators. In the core of a LWR, primarily coolant flows upward but their also exist a flow in lateral direction, called crossflow, in the gaps in between the sub-channels. The radial transport of scalars and vapors (i.e., crossflow mixing) due to crossflow has an important bearing on the assessment of the probable fuel rods damage. Motivated by this safety aspect of a LWR, the aim of the present study is to enhance the current understanding of crossflow mixing in a vertical tube and to develop a mixing model. An experimental study, supported by a few numerical simulations, was performed to identify different fluid flow patterns in a vertical tube bundle/similar geometry and the contribution of these patterns towards the lateral inter channel transport of a passive scalar. The experiments were performed in water at isothermal, single-phase flow and ambient operating conditions for Reynolds numbers ranging from approximately 900 to 22,000. The experimental results show the existence of counter rotating coherent vortices adjacent to the channel-gap interface for all flow regimes, i.e., laminar, transitional and turbulent flows. These coherent vortices are identified as a major contributor towards the crossflow. Experiments on the radial mixing of a passive scalar induced by crossflow show that the coherent structures impart a significant contribution towards crossflow mixing for all Reynolds numbers. Based on the experimental results a continuous stirred tank (CST) based crossflow mixing model has been proposed for the full range of channel Reynolds numbers. The model predictions are found to be within a band of ±20% of the experimental values for the full range of channel Reynolds numbers.