Powder Stream Research Articles

Integrated framework of multi−physics mechanism models for gas−powder flow, molten pool evolution and part deformation in high alloy steel additive manufacturing process

Multi−physics numerical models for metal additive manufacturing (MAM) could replace the expensive trial−and−error experiments, reveal the mechanisms of printing process, and provide an effective means of optimizing process parameters and mitigating part defects. However, the existing simulations of MAM process seldom consider the coupling effects between multiple recursive processes. There is an urgent need to overcome the dilemma of limiting MAM process simulation to a localized single sub−process. For directed energy deposition (DED) additive manufacturing process, the integrated simulation framework proposed in this study consists of a gas−powder flow model, a heat and fluid flow model, and a sequentially coupled thermal–mechanical model, which can be used to predict the feedstock feeding process, the evolution of molten pool, and the residual deformations of metal part, respectively. The accurate results of the three recursive sub−processes can be obtained by delivering the validated parameters between models. Corresponding experiments are conducted to verify the accuracy of the integrated framework based on 12CrNi2 alloy powders. The deviation of the predicted focal distance of powder streams is 5.14 %, and the deviations of the predicted molten pool height and width are all less than 8.5 %. The integrated framework could rapidly and accurately predict the residual deformation tendency and the position where the maximum deformation occurs. For single−track and eight−layer line deposition, the deviations of the predicted maximum deformation are all within 0.1 mm. The integrated framework could efficiently provide accurate and comprehensive solutions for optimizing DED process.

Influence of the inner roughness of powder channels on the powder propagation behavior in laser metal deposition

The powder propagation behavior of powder nozzles for the laser metal deposition process has a significant influence on powder utilization rate and track geometry. A well-focused powder stream will lead to a higher process efficiency and lower material loss. Powder channels with different roughness and a constant diameter of 1.5 mm were placed by wire electrical discharge machining into a copper alloy printed by powder bed fusion. Nickel base powder with a size of −106 to +45 μm was delivered through the powder channels with varied carrier gas flow rates and varied powder mass flow rates. High-speed imaging was used to analyze the powder flow. From these recordings, the dispersion angle of the powder stream from single channels could be measured as well as the velocity of particles. Moreover, the relationship between individual particle velocity and individual particle flight angle was investigated. It was found that the inner roughness of powder channels has a major impact on powder propagation behavior. It could be shown that with a decrease in Ra from 2.16 to 0.27 μm the divergence angle decreased by around 61% while the particle velocity was increased by at least 28% for all varied parameters. Particles with a high velocity tend to have a lower particle flight angle.

Investigating laser beam shadowing and powder particle dynamics in directed energy deposition through high-fidelity modelling and high-speed imaging

Laser beam shadowing in directed energy deposition process occurs when in-flight powder particles partially block the laser beam, preventing it from fully reaching the melt pool. This blockage alters the beam's profile, introducing spatio-temporal randomness in heat transfer to the melt pool and affecting its stability. This study utilizes a novel high-fidelity multiphysics model to quantitatively analyze gas-powder stream dynamics and laser-particle interactions. It specifically focuses on dynamic variations in laser beam shadowing, attenuation, and reflection of the laser beam radiant energy, along with powder particles temperature prediction from in-flight heating, accounting for multiple reflection phenomena. Gas-particle dynamics is modeled using a fully coupled CFD-DEM approach, while laser-particle in-flight interactions are predicted using a ray-tracing-based photon discretization method. Modeling predicted that interactions between the laser beam and the in-flight powder stream could result in energy losses up to approximately 70 % due to shadowing and attenuation. The study comprehensively explored the spatio-temporal variations in the attenuated laser beam profile, revealing significant deviations from its initial Gaussian profile and substantial stochasticity and randomness over time. To confirm the modeling results and correlate computational predictions with experimental data, high-speed visible imaging of the powder stream, both with and without the melt pool, along with laser beam attenuation measurements using a photodiode system, were conducted. The model's predictions of particle velocity and laser beam attenuation characteristics aligned with the results from these experiments. Modeling and high-speed imaging demonstrated that in-flight heating caused some particles to melt and few others to reach the vaporization threshold. High-speed imaging further validated these numerical findings by examining the interaction of heated particles with the melt pool across different phases (solid, liquid, and partially vaporized). A significant observation was the increased velocities of partially vaporized particles due to recoil pressure from selective vaporization, which led to the formation of deep craters in the melt pool resulting from high-speed impact.

Design of laser-powder coupling for high-speed laser direct energy deposition

High-speed laser direct energy deposition (DED-LB) is developed to enhance deposition efficiency while maintaining build quality, compared to conventional DED-LB. The core of designing high-speed DED-LB involves formulating the laser-powder coupling, through optimizing nozzle structural and deposition process parameters. This study developed an analytical model to delineate the coupling mechanisms between laser beam and discontinuous coaxial powder stream (DCPS) comprised of individual single powder streams (SPSs). A data pretreatment method, involving gray processing, normalization, and renormalization of the thermal images successively, was applied to validate the model. Simulation results indicate that the DCPS spot size and SPS particle temperature distribution are predominantly influenced, respectively, by the nozzle structural and deposition process parameters. Through jointly designing the nozzle structural and deposition process parameters, a high deposition rate of high-speed DED-LB, which is 1.7 times that of conventional DED-LB, was achieved. This paper pioneers a new design methodology for high-speed DED-LB.

The effect of the laser beam intensity profile in laser-based directed energy deposition: A high-fidelity thermal-fluid modeling approach

Modeling the thermal and fluid flow fields in laser-based directed energy deposition (DED-LB) is crucial for understanding process behavior and ensuring part quality. However, existing models often fail to accurately predict these fields due to simplifying assumptions, particularly regarding powder particle-induced attenuation in laser power and energy density distribution, and the variable material properties and process parameters. The present work introduces a high-fidelity multi-phase thermal-fluid model driven by a combination of the discrete element method (DEM) and the finite volume method (FVM). Incorporating an enhanced attenuation model for laser energy enables a more precise approximation of powder particle-induced attenuation effects in the laser power and energy density distribution. The study focuses on the influence of laser beam intensity profiles during DED-LB of austenitic stainless steel (AISI 316 L), with model validation conducted through experimental measurements of deposited track dimensions for different beam shapes. The results of numerical simulations demonstrate the critical impact of powder-induced attenuation on the laser power and intensity profiles. Neglecting laser energy attenuation, a common assumption in numerical simulations of DED-LB, leads to overestimations of the absorbed energy of the laser beam, affecting thermal and fluid flow fields, and melt pool dimensions. The present study unravels the complex relationship between the attenuation coefficient (due to the powder stream) and powder stream characteristics, describing the variations of the attenuation coefficient with changes in the powder mass flow rate and powder stream incidence angle. The findings show the critical effects of laser beam shaping on melt pool behavior in DED-LB, with square beams inducing larger melt pool volumes and circular beams creating smaller but deeper melt pools. The proposed enhanced thermal-fluid modeling framework offers a robust approach for optimizing laser-based additive manufacturing across diverse materials and laser systems.

Influence of the laser-beam intensity distribution on the performance of directed energy deposition of an axially fed metal powder

This paper investigates the influence of laser-beam intensity distribution (LBID) on the performance of the annular laser-beam directed energy deposition (DED) process with axial powder delivery. Three different LBIDs: Gaussian-like (G-LBID), top-hat-like (TH-LBID) and ring (R-LBID) at two LBID diameters were used. The process performance was characterised qualitatively in terms of the melt-pool shape and the process stability and quantitatively by powder-catchment efficiency, selected geometrical and metallurgical properties of the clad. The observed influence of LBID on process performance, as determined by the relationship between LBID and powder stream density distribution (PSDD), decreased with increasing mean surface-energy density and was more significant at larger LBID diameter. The highest powder-catchment efficiencies (90% and 87%) were achieved with the G-LBID and TH-LBID, whose high-intensity centre is aligned with the peak of the Gaussian-like PSDD. A lower powder-catchment efficiency of 77% was achieved with the R-LBID, whose high-intensity region is located at the edge of the melt pool with minimum powder density. However, this also results in the highest and most uniform dilution, the highest metallurgical bond ratio and the lowest lack of fusion porosity at the clad-substrate interface. In addition, the process was stable at the lower values of mean surface-energy density with R-LBID, while balling instability was observed with G-LBID and TH-LBID. It can be concluded that the use of R-LBID at lower values of mean surface-energy density improves the performance of the DED process with axial powder feed in terms of process stability and metallurgical properties of the clad.

Laser metal deposition of titanium on stainless steel with high powder flowrate for high interfacial strength

Direct joining of titanium and stainless steel 316 L with a strong interface is very challenging due to the formation of the brittle intermetallic compounds FeTi and Fe2Ti in the intermixing zones and to the high residual stress induced by the mismatch of the thermal expansion coefficients. In this bimetallic directed energy deposition study, firstly, deposition of Ti on stainless steel was carried out using conventional process parameter regime to understand the interfacial cracking susceptibility and then a novel high powder flowrate approach is proposed for controlling the dilution and constraining the intermetallic phases forming at the interface. The influence of high temperature substrate preheating (520 °C) on the cracking susceptibility and interface strength was also investigated. The deposited Ti samples and their interfaces with the 316 L substrate were characterized with optical microscopy, scanning electron microscopy and energy dispersive X-ray spectroscopy to investigate the geometry, microstructures and chemical compositions in relation to the cracks. The high powder flowrate deposition of Ti on stainless steel 316 L results in an extremely thin dilution region (~ 10 μm melt pool depth in the substrate) restricting the formation of the intermetallic phases and cracks. The ultimate shear strength of the interfaces of the crack free sample was measured from cuboid deposits and the highest measured strength is 381 ± 24 MPa, exceeding the weaker base material pure Ti. The high interfacial strength for high powder flowrate deposition is due to the substantial attenuation and shadowing of the laser beam by the in-flight powder stream as demonstrated by the high-speed imaging resulting in an extremely small dilution region.

Relevance of inter-particle interaction in directed energy deposition powder stream

Blown powder dynamics is one of the aspects of Directed Energy Deposition (DED) with a major influence on deposition quality and potential for improvement in simulation. Most currently employed computational models discard powder grain collisions as negligible, although little explicit evidence for that claim exists. A Discrete Element Method approach is thus employed in the present work to simulate the actual number of grain collisions in a powder stream of a commercial discrete coaxial nozzle and how that number varies with the key processing parameters. While the number of powder grain collisions is found to in fact be negligible at one side of the usable parameter spectrum, the opposite side results in as many as 84% of all the powder grains being involved in inter-granular collisions with an average rebound angle of 14°, challenging the established hypothesis of the negligibility of this phenomena.

Adaptive powder nozzle setup for enhanced efficiency in laser metal deposition

Laser metal deposition (LMD) is a blown powder process used for the additive manufacturing of large and/or complex parts. The laser spot size is determined by the fiber optic cable and the imaging ratio of the process optics. Spot sizes typically used in LMD can range from 200 μm to several millimeters, whereby zoom optics can be employed to change the laser spot focus within seconds during the process. However, industrial powder nozzles are still static in terms of powder spot size. Changing the powder spot size in line with the laser spot size could ensure the favorable dual outcome of time savings when printing large volumes while also generating fine near-net-shape features. To help overcome the current limitations in the LMD process, this work examines an adaptive powder nozzle setup. In this discrete coaxial layout of three single lateral powder injectors, the individual powder injectors can be adjusted closer to or further from the process to, respectively, dilate or shrink the powder stream focus. Different inner diameters of powder injectors are hereby examined. The resulting powder propagation behavior is characterized for different setups of the single powder nozzles. Single beads are welded with different nozzle setups for fine and coarse powder spots, while the laser spot size is changed accordingly using zoom optics. The laser power is a closed-loop controlled by a two-color pyrometer to achieve comparative process temperatures. The single beads are evaluated with regard to their geometry. High-speed imaging provides supplementary information on weld bead generation.

Investigation of thermal behavior of powder stream and molten pool during laser-based directed energy deposition

Laser-powder interaction and molten pool evolution are two main physical processes in laser-based directed energy deposition (L-DED), which greatly affect the morphology and quality of the deposition layer. In this research, a comprehensive three-dimensional analytical-numerical model which integrates a laser-powder interaction model and material deposition model is developed to study the multi-physics coupling characteristics in L-DED process. The concentration of the powder stream is modeled based on conservation of mass and is in good agreement with the experimental result which is captured by high-speed camera. The effects of main parameters, including laser power, powder feeding rate, powder feeding angle and defocus length on the laser attenuation and the temperature distribution of the in-flight powder particles are analyzed. The results show that the laser attenuation rate increases from 0.72% to 1.36% with the powder feeding angle increasing from 45 deg. to 65 deg-. However, the laser attenuation rate is almost the same with the defocus length of 25 mm, 35 mm and 45 mm. Moreover, the peak powder temperature on the deposition surface increases with the increase of powder feeding angle and decrease of defocus length. Accordingly, the heat and mass input conditions at the molten pool surface for the material deposition model, including powder mass flux, effective laser intensity and powder temperature are obtained from the laser-powder interaction model. The heat transport and fluid dynamics of the molten pool are discussed based on the calculated results from material deposition model. Finally, the calculated molten pool geometry shows good agreement with the experimental results with the relative error <8.5%. This work is helpful in process optimization from the point of view of adjusting the parameters related to the powder stream and deeper understanding of laser-powder interaction and molten pool evolution during the L-DED process.

Image Based In-Process Powder Stream Monitoring during Laser Metal Deposition for the Measurement of the Gravimetric Powder Mass Flow Rate and the Detection of Deviations in the Powder Stream

In laser metal deposition, the quality of the applied clad as well as the additively manufactured component is directly linked to the precise powder feeding into the laser-induced melt pool. Despite the highly automated and closely monitored process, there is currently no industrially established in-process system for the powder stream monitoring. In this work, the powder is illuminated with a line laser in a defined horizontal plane to visualize the back-reflecting light to the camera. Optical filtering is needed to attenuate the process emissions, which is adjusted to the wavelength of the illumination laser. The powder stream is continuously monitored by observing grayscale values. The gravimetric measurements of the powder mass flow are correlated to the grayscale values to display the powder mass flow in absolute mass flow values from the image data. This is qualified by the detection of defined and temporarily introduced deviations in the powder stream.

Design and Numerical Analysis of an Inside-Beam Powder Feeding Nozzle for Wide-Band Laser Cladding.

Wide-band laser cladding technology has emerged as a solution to the limitations of traditional cladding techniques, which are small single-path dimensions and low processing efficiency. The existing wide-band cladding technology presents challenges related to the high precision required for the laser-powder coupling and the significant powder-divergence phenomenon. Based on the inside-beam powder-feeding technology, a wide-band powder-feeding nozzle was designed using the multi-channel powder flow shaping method. The size of the powder spot obtained at the processing location can reach 40 mm × 3 mm. A computational fluid dynamics analysis using the FLUENT software was conducted to investigate the impact of the nozzle's structural parameters on the powder distribution. It was determined that the optimal configuration was achieved when the powder-feeding channel was 8, and the transverse and longitudinal dimensions for the collimating gas outlet were 0.5 mm and 1 mm, respectively. Among the process parameters, an increase in the carrier gas flow rate was found to effectively enhance the stability of powder transportation. However, the powder feed rate had minimal impact on the powder concentration distribution, and the collimating gas flow rate appeared to have a minimal effect on the divergence angle of the powder stream. Wide-band laser cladding experiments were conducted using the designed powder-feeding nozzle, and a single-path cladding with a width of 39.96 mm was finally obtained.

Continuous manufacturing of pharmaceutical products: A density-insensitive near infrared method for the in-line monitoring of continuous powder streams

Near infrared (NIR) spectroscopy is a valuable analytical technique for monitoring chemical composition of powder blends in continuous pharmaceutical processes. However, the variation in density captured by NIR during spectral collection of dynamic powder streams at different flow rates often reduces the performance and robustness of NIR models. To overcome this challenge, quantitative NIR measurements are commonly collected across all potential manufacturing conditions, including multiple flow rates to account for the physical variations. The utility of this approach is limited by the considerable quantity of resources required to run and analyze an extensive calibration design at variable flow rates in a continuous manufacturing (CM) process. It is hypothesized that the primary variation introduced to NIR spectra from changing flow rates is a change in the density of the powder from which NIR spectra are collected. In this work, powder stream density was used as an efficient surrogate for flow rate in developing a quantitative NIR method with enhanced robustness against process rate variation. A density design space of two process parameters was generated to determine the conditions required to encompass the apparent density and spectral variance from increases in process rate. This apparent density variance was included in calibration at a constant low flow rate to enable the development of a density-insensitive NIR quantitative model with limited consumption of materials. The density-insensitive NIR model demonstrated comparable prediction performance and flow rate robustness to a traditional NIR model including flow rate variation (“gold standard” model) when applied to monitoring drug content in continuous runs at varying flow rates. The proposed platform for the development of in-line density-insensitive NIR methods is expected to facilitate robust analytical model performance across variable continuous manufacturing production scales while improving the material efficiency over traditional robust modeling approaches for calibration development.

Laser cladding: A high-speed-imaging examination of powder catchment efficiency as a function of the melt pool geometry and its position under the powder stream

This paper provides quantitative information about the paths taken by blown powder particles during laser cladding. A proportion of the powder is “wasted” by bouncing off the solid areas surrounding the melt pool. This wastage reduces the productivity and profitability of the process. In this paper, specially developed software was used to analyze high-speed imaging videos of the cladding process, to monitor the directions of powder particle flight toward and away from the melt pool area. This information has been correlated to the geometry and position of the melt pool zone for three different cladding techniques: single track cladding (A tracks), standard overlapping track cladding (AAA cladding), and a recently developed technique called ABA cladding. The results show that the melt pool geometry, and particularly the overlap between the melt pool and the incoming powder stream, has a strong influence on powder catchment efficiency. ABA cladding was found to have considerably better powder catchment efficiency than standard AAA cladding and this improvement can be explained by consideration of the geometries and positions of the melt pools and surrounding solid material in each case. As powder costs are an important factor in industrial laser cladding, the adaption of the ABA technique, and/or control of pool/powder stream overlap (e.g., by making the powder stream not coaxial with the laser beam), could improve the profitability of the process.

Dynamics of the laser–powder interaction in the Ti6Al4V powder feeding process of laser-directed energy deposition additive manufacturing

Laser-directed energy deposition (L-DED) utilizes a high-power laser and coaxial powder feed to rapidly fabricate high-quality components with intricate structures. However, the complex interaction between the powder stream and laser beam is not yet fully understood. Particularly, the morphological evolution of the droplets after powder melting is closely related to the quality of component formation. In this study, the dynamical mechanism of the aggregation and evolution of molten powder after laser–powder interaction was investigated. The Ti6Al4V powder flow was observed at various laser powers using a high-speed camera. Imaging of the molten droplets at approximately 50,000 frames·s−1 was conducted, depicting the aggregation of molten powder into droplets of different sizes and shapes, mainly consisting of spherical and ribbon droplets, after laser–powder interaction. A transient numerical model was established with consideration of the mutual influence of gravity, recoil pressure, and surface tension, depicting the temporal evolution of the temperature field and droplet morphology. The simulation results indicated that the balance between the surface tension, gravity, and recoil pressure determined the morphological evolution of the molten droplets. When spherical and ribbon droplets fell into the molten pool, the maximum velocities within the pool increased by 24 % and 69 %, respectively. This phenomenon reduced the stability of the molten pool to a certain extent and affected the print quality of the part. This study deepens our comprehension of the dynamic mechanisms of Ti6Al4V alloy powder melting evolution, providing invaluable insights for optimizing L-DED processes.

Melt pool dynamics on different substrate materials in high-speed laser directed energy deposition process

High-speed laser directed energy deposition (HSL-DED) is a variant of the laser directed energy deposition process where a defocused metal powder stream is used, and it typically involves processing speeds exceeding 5 m/min. However, the interactions between the laser beam, powder stream, and substrate surface in HSL-DED have not been extensively studied. This study used a specialized XIRIS XVC-1000 welding camera with a narrow bandpass filter to record the interaction phenomenon. These observations were first carried out without powder delivery, using laser surface melting techniques, and involved processing speeds of up to 20 m/min and laser powers of up to 3 kW. HSL-DED with powder delivery was then conducted with the same parameter combinations for comparative analysis. The in situ observations in laser surface melting and HSL-DED identified a physical separation between the laser spot and the melt pool boundary, referred to as melt pool lag. Different substrates’ chemical compositions and the resulting thermophysical properties significantly impact melt pool dynamics during the high-speed laser-material interactions for a given process condition. The findings from this work have enabled a better understanding and control of melt pool dynamics in HSL-DED.

Characterization of optical emissions during laser metal deposition for the implementation of an in-process powder stream monitoring

In laser metal deposition (LMD), the powder is fed into the laser-induced melt pool using different powder nozzles for the purpose of additive manufacturing and the generation of wear and corrosion protection coatings. So far, there are no industrially established in-process monitoring systems for the powder stream but mainly measuring systems that examine the powder stream propagation offline and without the processing laser. A challenge in implementing an image-based in-process monitoring system is the process illumination for the distinction of the powder particles from the background radiation caused by the processing laser and the melt pool. To overcome this challenge, filtering is needed to attenuate the process emissions and simultaneously brighten the powder stream. Therefore, this work focuses on generating a continuous high contrast between the powder and the background. The powder particles are illuminated by a light source mounted laterally to the powder stream in the horizontal plane below the nozzle opening to make the reflecting powder particles visible to the camera. The optical process emissions were characterized during LMD with respect to the influence of an increasing laser power, which was presented in correlation to the increasing process emissions. The evaluation of the spectrograms has made it possible, due to the adapted illumination and filtering, to ensure a constantly high contrast between the process emissions and the powder so that online monitoring of the powder stream was implemented successfully during the LMD process despite the active processing laser.

Prediction of the bead profile for wide-band laser cladding via hybrid model of mathematics and thermal simulation

Wide-band laser cladding has high deposition rate and low dilution due to the large laser spot and rectangular powder nozzle. A mathematical model was proposed to predict the track cross-section profile for wide-band laser cladding. The melt pool morphology and powder spatial concentration were taken into consideration to improve the accuracy of the model. Thermal field simulation was carried out to study the melt pool evaluation with process parameters, while infrared images were used to verify the simulation fidelity. Powder stream photographs were taken via a CCD (charge coupled device) camera to obtain the powder spot size. Single and multi-track cladding experiments were conducted, and calculated contours of the model coincide well with the experimental results. Simulated and measured melt pool width and length agree with in 2.16% and 2.36% of error, respectively. The error of predicted cross section area is less than 9.35%. The model also has good ability to predict the powder efficiency and morphology of the cladding layer.

Scaling laws and numerical modelling of the laser direct energy deposition

A computational framework that can be easily implemented into a two-phase flow model has been developed for DED (direction energy deposition) process. The information on the in-flight powder stream is treated by an analytical model and regarded as input to the cladding model. The cladding model utilizes Computational Fluid Dynamics (CFD) technique with the volume of fraction (VOF) method, which considers interfacial dynamics and interaction with the powder stream.Furthermore, we have applied Buckingham π theorem to develop scaling laws for the geometry of the final track. In contrast with the dimensional empirical models proposed in the previous studies, the dimensionless laws in this study can largely match the trends of measured data with noticeable variations of working conditions collected from several different studies. The dimensionless cladding width is dominated by the dimensionless laser power with a positive correlation. The increase in dimensionless laser power and powder mass flow rate usually results in a greater value of the dimensionless cladding height, but this trend could become opposite when the values of catchment and energy efficiencies are close to 100%. A threshold that dilution suddenly goes up has also been provided and validated based on the dimensionless parameters. This study provides a reliable numerical model and scaling laws with sufficient generality. The processing window has also been carried out with constraints of dilution and contact angle, bringing design guidelines to the related industries.

Influence of powder stream density on near infrared measurements upon scale-up of a simulated continuous process

Near-infrared (NIR) spectroscopy is a powerful process analytical tool for monitoring chemical constituents in continuous pharmaceutical processes. However, the density variation introduced when quantitative NIR measurements are performed on powder streams at different flow rates is a potential source of a lack of model robustness. Since different flow rates are often required to meet the production requirements (e.g., during scale-up) of a continuous process, the development of efficient strategies to characterize, understand, and mitigate the impact of powder density on NIR measurements is highly desirable. This study focused on assessing the effect of powder physical variation on NIR by enabling the in-line characterization of powder stream density in a simulated continuous system. The in-line measurements of powder stream density were facilitated through a unique analytical interface to a flowing process. Powder streams delivered at various design levels of flow rate and tube angle were monitored simultaneously by NIR diffuse reflectance spectroscopy, live imaging, and dynamic mass characterization. Statistical analysis and multivariate modeling confirmed powder density as a significant source of spectral variability due to flow rate. Besides providing broader process understanding, results elucidated potential mitigation strategies to facilitate effective continuous process scale-up while ensuring NIR model robustness against density.