Effect of welding heat input conditions on the dynamic behavior of pulse laser beam welding molten pool for Ti6Al4V thin plate with clearance

For a practical pulse laser beam welding (PLBW) process of metal sheets assembled in butt joint configuration, the precise control of the assembling clearance has been a challenge. The existence of machining burrs and assembly errors will lead to forming severe defects, such as misalignment, welding leakage, and penetrating. In this paper, a pair of Ti6Al4V plates with a 0.2-mm air gap was tested by an improved PLBW process. A three-dimensional multi-phase and multi-physical field coupling model of the Ti6Al4V plate with a reserved air gap was established according to the weld profile, and the dynamic behavior of the keyhole and molten pool was simulated. The transient temperature field, velocity field, keyhole size, and liquid bridge connection were calculated by using different welding heat input parameters. The results showed that the weld profile simulated by the CFD model is in good agreement with the experimental results, and the deviation is between 0.68 and 7.95%. After the laser power reaches the peak value, the metal steam eruption weakens and the obvious Marangoni vortex appears in the molten pool. The simulated keyhole is always in three stages, that is, the keyhole appears and then gradually forms the through-hole. The through-hole keeps oscillating, and finally, the keyhole shrinks and disappears when the laser power drops to zero. With the increase of laser peak power, the keyhole shape becomes more curved, indicating that the keyhole oscillation is enhanced. With the increase in welding speed, the stability of the molten pool is improved, and the area of the liquid bridge rises more regularly.


Introduction
As we all know, titanium alloy is widely used in aerospace, medical field, material field, and petrochemical field. Ti6Al4V is the most commonly used among all titanium alloys because of its low density, high specific strength, and corrosion resistance [1]. However, it has the disadvantages of high melting temperature, low thermal conductivity, and easy oxidation during processing, which is not compatible with the traditional process [2,3]. The inherent material characteristics of Ti6Al4V put forward stricter requirements on welding conditions. Among many welding technologies, laser welding recently gained popularity as a promising joining technology for Ti6Al4V in industries because of its high weld quality, high precision, high efficiency, excellent flexibility, and low deformation. According to the consistency of the output energy, laser beam welding can be split into two types: pulse laser beam welding (PLBW) and continuous laser beam welding (CLBW). The PLBW can be thought of as a combination of numerous single-spot welding. As shown in Fig. 1, the laser irradiates the metal material for a specified period in each cycle of a pulsed laser action, and then, the laser closes the material for cooling [4].
In recent years, a lot of research has been done on the PLBW of titanium alloy. The microstructure, mechanical properties, and fracture mechanism of welded joints have been discussed. Mohid et al. [5] showed that the pulse energy was the key determinant for welding penetration, while the pulse width highly influenced the weld pool stability and weld bead homogeneity. Junaid et al. [6] used pulsed Nd:YAG laser beam welding and pulsed tungsten inert gas welding to weld 1.6-mm-thick Ti-5Al-2.5Sn α titanium alloy sheet. Less residual stresses, deformation and distortion, and superior mechanical properties in PLBW made the process more feasible than pulsed tungsten inert gas welding for Ti-5Al-2.5Sn alloy sheets. Gao et al. [7] studied the effect of weld section and microstructure on mechanical properties of pulsed Nd:YAG laser-welded Ti6Al4V alloy under different welding conditions. Kumar et al. [8] conducted experimental and numerical analysis on Nd:YAG laser welded titanium alloy sheet temperature distribution. They found that the elongated grain and regular plate structure in Nd:YAG laser-welded joint are the key factors in improving the joint's tensile strength.
For a practical PLBW process of metal sheets assembled in butt joint configuration, the precise control of the assembling clearance has long been a challenge. The existence of machining burrs and assembly errors will lead to forming severe defects, such as misalignment, welding leakage, and penetrating [9]. The research mentioned above on PLBW of titanium alloy shows that PLBW is an exemplary process for titanium alloy welding. However, many studies did not consider the impact of assembling clearance. The existence of assembly clearance increases the complexity of the welding process and affects the morphology of the welding pool and the flow of liquid metal. It is difficult to obtain the temperature distribution of molten pool and keyhole evolution accurately by conventional process experiments. Numerical modeling and simulation can provide insight into the welding process and help to reveal the mechanism of process parameters. In recent years, the numerical simulation of the PLBW process has made significant progress. Satyanarayana et al. [10] built a three-dimensional heat-flow coupling model to simulate heat transfer and Marangoni flow in pulsed Nd:YAG laser welding of Zr-Nb alloy. The calculation of heating and cooling rates facilitated the explanation of microstructure formation in various seam zones. Pang et al. [11] proposed a quantitative model of porosity defects induced by keyhole instability in a titanium alloy's partial penetration CO 2 laser welding. The simulated keyhole depth fluctuations could represent the various trends in pore number and average size for the studied process conditions. Drobniak et al. [12] used multi-physical CFD modeling to study the process of two stainless steel plates with overlapping configurations. The influence of clearance on fluid dynamics and weld defect appearance was analyzed. Based on the results, adapted beam shapes were proposed that allow for optimized energy distribution and thus an improved weld quality. Hao et al. [13] established a threedimensional transient numerical analysis model of tungsten inert gas welding with reserved clearance. The model takes into account the loss of total input heat as well as the change in arc pressure produced by the reserved clearance. The modeling findings suggest that increasing the reserved clearance increases the maximum sag of the molten pool.
To sum up, the existence of assembly clearance has a significant impact on the heat transfer and fluid flow in the molten pool. The numerical investigations reported have paid little attention to keyhole and weld pool dynamics modeling and calculation in pulse laser welding of titanium alloys with reserved air gaps. The influence of crucial heat input parameters on heat transfer, keyhole evolution, and fluid flow in the pulse laser tailor-welding process has not been fully solved. Here, we develop a three-dimensional multi-phase and multi-physical field coupling model to simulate the PLBW process of Ti6Al4V plates with reserved air gaps, involving the dynamic behavior of keyholes and molten pools. THINC/LS algorithm and ray tracing algorithm are used to realize free interface tracking and energy calculation on the interface. Using various welding heat input parameters, the transient temperature field, velocity vector, keyhole size, and liquid bridge area were calculated and examined. The effects of welding speed and laser peak power on heat and mass transfer behavior in the PLBW process were investigated.

Experimental details
Ti6Al4V plates were used in the experiment, with dimensions of 100 mm × 50 mm × 1.2 mm. Ti6Al4V belongs to the α + β type dual-phase titanium alloy. Its main alloying elements are Al and V, which chemical composition is shown in Table 1. Before welding, the specimens were assembled in a butt way, and a 0.2-mm air gap was reserved. The establishment of the experimental system is shown in Fig. 2.
The welding equipment is a Trumpf Trudisc 4001 disc laser with a wavelength of 1.03 μm and a maximum laser power of 4 kW. The laser waveform is modulated as shown in Fig. 3, and the spot diameter is 0.7 mm. Pure argon was supplied to the top and bottom surfaces of the sample, with Welding process parameters are shown in Table 2. The influence of different heat input conditions on the dynamic behavior of the keyhole and weld pool was studied by changing the welding speed and peak power. After welding, metallographic samples are prepared according to standard procedures. The weld cross section was obtained by wire cutting the middle of the specimen and then ground and polished to 1-μm diamond paste. The metallographic sample of the weld section was etched with Kroll solution (2% HF + 10% HNO 3 + 88% H 2 O) and observed by a 4XC metallographic microscope.

Modeling simplification
The critical physical factors in the PLBW process of thin plate with reserved clearance are not only basic convection in the molten pool to transfer heat but also critical physical processes that affect the molten pool and keyhole change, such as multiple beam reflections, Rayleigh scattering, Fresnel absorption, and solid-liquid-gas phase transitions. To ensure accuracy and improve the calculation efficiency, the following assumptions are made: 1. The computational domain consists of incompressible, Newtonian, and laminar fluids. 2. The original laser beam consists of a limited number of sub-beams, which irradiate vertically on the surface of the workpiece.  3. The reflection of the sub-beam at the liquid-gas interface follows the principle of specular reflection. 4. Due to the weak ionization plume effect, the inverse bremsstrahlung absorption can be neglected [14]. 5. In view of the weak ionization of metal vapor in laser welding, the mixture of metal vapor and shielding gas argon is simplified to a single argon phase [15]. 6. An additional surface heat source is used to equivalent the plume heating effect around the keyhole opening.

Governing equations
For incompressible fluids, the mass conservation is as follows [16]: where � ⃗ u is the velocity field of fluid. The momentum equation considers the momentum source terms, including gravity, heat buoyancy, Darcy resistance in solidification and melting regions, and recoil pressure [17].
Here, t, ρ, p, , f g , f b , f d , and f r denote time, mass density, pressure, dynamic viscosity, gravity, heat buoyancy, Darcy resistance, and recoil pressure, respectively.
In Eq. (5), and T l are thermal expansion coefficient and liquidus temperature. In Eq. (6), l is liquid volume fraction, c is usually 10 −3 to 10 −5 . K 0 is Darcy constant related to the volume fraction of the liquid phase which in this paper is valued at 2 × 10 7 . In Eq. (6), A and B 0 denote pressure constant and evaporation constant. M a , L v , N a , k b , and � ⃗ n are molar mass, evaporation latent heat, Avogadro constant, Boltzmann constant, and surface normal vector, respectively.
To improve the numerical calculation speed, the liquid fraction l changes linearly with temperature: with T s being solidus temperature and T 0 being reference temperature.
Considering the energy balance of laser heat flow, the pulse laser welding process of the thin plate should follow the energy conservation equation [18]: Here, h is enthalpy and k is heat conductivity. S Q is the energy source term that consists of Rayleigh scattering Q Rs and Fresnel absorption Q Fa generated during the continuous laser output.

Free surface tracking algorithm
The volume of fluid (VOF) [19] and level set method (LS) [20,21] have been widely used to solve the two-phase flow problem. In this paper, the THINC/LS algorithm reported by Qian et al. [22] is improved to solve the keyhole evolution under laser-induced evaporation and oscillation. The calculation accuracy of interface geometric features can be considerably improved with this method, which combines the THINC and LS functions. In a vector space, the Level Set approach defines a symbolic distance function. The Level Set function has the following attributes: The THINC function uses the hyperbolic tangent function to fit the interface discontinuity solution within the range of [0,1]. When its value is 0.5, it represents the moving interface, which reads Here, is a parameter used to control the thickness of the interface. The Heaviside step function can be approximated by increasing the value of . The average value of the THINC function H represents the volume fraction of the fluid specified in the VOF method.
where Ω denotes the computing domain.
The evolution of the keyhole surface can be described by interface moving functions: According to Eq. (10) and Eq. (11), when (x, t) = 0 or H(x, t) = 0.5 indicates the moving interface. THINC/LS method can update the LS function and THINC function at the same time to achieve high-order interface representation and numerical conservatism.

Heat source model
In the PLBW process, the ray-tracing algorithm was created for heat source modeling. Beam discretization, beaminterface intersection, multi-reflections, Rayleigh scattering, and Fresnel absorption are all part of the ray-tracing process [23]. The laser spot is discretized into a finite number of beams using a two-dimensional uniform grid, and the laser power density distribution is ideally Gaussian. The incident sub-beam advances gradually along the unit vector, and the energy attenuation caused by Rayleigh scattering Q Rs can be calculated along the laser propagation path. When the free interface meets the laser beam, the Fresnel absorption Q Fa can be calculated. Q Rs and Q Fa are calculated as follows [18]: Here, l , k r , m , mr , n , and bn are the beam propagation distance after divergence, Rayleigh scattering coefficient, multi-reflection times, total number of multiple reflections, sub-beam number, and total number of sub-beams, respectively. ⃗ I and � ⃗ n t are unit incident vector and unit normal vector. Rs and Fa are the coefficients related to Rayleigh scattering attenuation and Fresnel absorption, respectively. Rs and Fa are calculated as follows [18]: In Eq. (16), r p , , m , and N denote vapor particle radius, laser wavelength, complex refractive index, and particle density, respectively. In Eq. (17), is the reflection angle of the sub-beam. According to the Hagen-Rubens relationship, Fa is highly dependent on the reflection angle.
Under the action of laser irradiation, the metal melts and evaporates to form a keyhole, forming a high-energy plasma above the molten pool. The lack of plume heating in the modeling will lead to the mismatch between the simulated molten pool width and the experimental molten pool width. As suggested by Yang et al. [24], two additional surface heat sources are applied on the top and bottom surfaces of the workpiece to equivalent plume heating effect. And the heat flux satisfies the following equation: Here, k 0 , P L , and t are heat source efficiency, laser power, and time, with r 1 and r 2 representing the heat source radius of surfaces and 1 and 2 representing the absorption coefficients. The radius of the surface heat source matches the actual weld size, so as to make the two surface heat sources conform to the actual working conditions.

Model development
As shown in Fig. 4, a computation domain of 5 mm × 2.2 mm × 1.5 mm is established and the mesh model consists of 132,000 cube elements with a grid step of 0.05 mm. The domain is initialized to the liquid phase region and the gas phase region. The initial liquid phase zone is filled with Ti6Al4V titanium alloy and separated (15) by a reserved air gap of 0.2 mm, while the gas phase region is filled with argon. The thermos-physical properties of Ti6Al4V alloy and argon are shown in Table 3 [25]. The top and bottom boundaries of the domain are set to pressure outlet and adiabatic. The boundary conditions of the walls in terms of energy are convective heat transfer and thermal radiation [26], which reads the following: Here, , h c , , and T 0 are surface emissivity, Stefan-Boltzmann constant, convective heat transfer coefficient, and reference temperature.

Simulation implementation
The welding process parameters applied in the simulation, such as laser spot diameter, laser peak power, welding speed, and other related parameters are highly consistent with the experimental parameters. The numerical model including calculation domain, three-dimensional mesh, and Ti6Al4V thermo-physical property parameters was loaded into fluent 19.0. The momentum and energy source terms are coded into a user-defined function (UDF) in C language. The pressure implicit algorithm is used to solve the governing equations, which are discretized using the finite volume approach. The PLBW process was simulated for 100 ms with a physical time step of 0.001 ms.

Verification of simulation results with weld cross section
As shown in Fig. 5, well-formed and defect-free weld beams composed of continuously overlapping weld spots can be obtained by using different welding speeds and laser peak power. As demonstrated in Fig. 6a, the simulated weld bead consisting of two weld spots is validated  using the experimental determined one. The simulated weld pool shape is compared with the weld cross section determined by the experiment, and the measured important feature dimensions are listed in Table 4. The calculated surface width (SW), half-penetration width (HW), and root width (RW) are 0.50-7.39% smaller than the experimental results. The primary cause of this error is that the secondary heating effect of the metal vapor plume around the keyhole opening is not taken into account.

Evolution characteristics of temperature field, velocity field, and liquid bridge profile
Using case 3 as an example, Fig. 7 depicts the evolution of the molten pool temperature field and phase interface in the first laser pulse period. The workpiece is colored by the temperature legend of 400 ~ 2600 K, and the solidus of the molten pool is drawn by the black line. At the beginning of the PLBW, the edge of the thin plate with a gap is heated to the evaporation temperature as the laser power is increasing. With the irradiation of the laser at t = 0.5 ms, two pits emerge on the edge of the sheet, but the two sheets are still separated at this moment. With more melting and penetration of the keyhole, the molten metal accumulated rapidly in the next 1 ms, and a liquid bridge was established in the gap.
In the period of 2 to 5 ms, the width and length of the molten pool increase with the increase of laser power. Looking in the Y direction, the opening at the bottom of the keyhole becomes larger, reflecting the slight changes in the shape of the keyhole. When the laser power reaches the maximum, the opening at the top of the keyhole is significantly larger than before, and the length and width of the top molten pool also reach their maximum. After t = 5 ms, the laser power is in the decline stage. With the decrease in temperature, the falling recoil pressure cannot balance the surface tension and hydrostatic pressure, resulting in the rebound of the keyhole surface from bottom to top and the backfilling of molten metal, which is consistent with the research reported by Chen et al. [18]. When the laser power drops to zero at t = 15 ms, the keyhole at the bottom is closed and disappears rapidly within microseconds. During the cooling process, the liquid bridge is further solidified into metal in the air gap, and the size of the molten pool is significantly reduced. The second laser pulse period has a calculation range of 25 to 50 ms. When the laser beam is re-irradiated from 25 to 26 ms, a portion of the solidified metal from the first pulse cycle is remelted and recoiled downward to form the back of the keyhole, as shown in Fig. 8. The molten metal forms a new liquid bridge on the front of the keyhole in the next microsecond. The length and diameter of the molten pool greatly grow over the 27-30-ms period, which is identical to the calculated results of the first pulse period. The top surface of the molten pool, on the other hand, resembles a droplet, which is a unique phenomenon of the second pulse period when compared to the first pulse period. During the remaining time, the dynamic behaviors of the molten pool such as temperature drop, keyhole rebound, volume shrinkage, and cooling occur due to the decrease in laser power. The keyhole oscillation and rebound lasted from 31 to 41 ms. At t = 40 ms, the keyhole is closed at the bottom. The keyhole rebound process not only causes the liquid bridges around the keyhole to merge but also makes the second welding joint partially overlap with the first welding joint. This overlap increases the length of the connection between the two welding joints.
When the pulsed laser beam is used to weld Ti6Al4V thin plate with a gap, the velocity field of the molten pool has the characteristics of metal vapor eruption and molten metal convection. The shift in velocity field around the keyhole is depicted in Fig. 9. The eruption speed of metal vapor is much higher than that of molten metal flow. The maximum eruption speed of metal vapor can reach 90.54 m/s, while the flow rate of liquid metal is lower than 2.0 m/s, which is consistent with the research results of Chang et al. [27]. The color arrow is used to show the flow velocity of molten metal due to the considerable difference between gas and liquid phase velocities. The arrow points in the same direction as the flow, and different colors represent varying flow velocities. The magnitude of the gas phase velocity is indicated by the length of the black arrow. The velocity vector of molten metal is perpendicular to the keyhole surface during laser heating and keyhole growth, as observed at 1.5 ms. The metal vapor eruption outside the keyhole becomes more powerful as the laser power increases. When t = 5.5 ms, the laser power reaches its maximum and then progressively falls, the metal vapor eruption weakens, and Marangoni convection is observed in the molten pool. The velocity of molten metal near the keyhole is larger than that at the molten pool's edge due to the temperature gradient. Under the operation of Marangoni shear stress, it can be observed that molten metal flows from the high-temperature keyhole area to the low-temperature area of the molten pool at t = 10 ms. The rebound pressure drops to zero during the  To further describe the Ti6Al4V thin plate connection process on the air gap, the regions with liquid volume fraction greater than 0.5 in the X-Y plane of z = 0 were extracted. The appearance of molten metal indicates that the thin plate has a connection on the gap, as presented at 1.5 ms in Fig. 10. The first liquid bridge is located in the middle of the thin plate's thickness direction, rather than on its upper and lower surfaces. The liquid bridge formed on the gap gradually expands as the welding time increases. The keyhole divides the liquid bridge into the front and rear parts, and there is no liquid bridge inside the keyhole. Under the influence of gravity, the molten metal travels downhill, causing visible liquid sag at the root surface of the workpiece. At the moment of 15 ms, the laser power drops to zero and the recoil pressure rapidly drops to zero. The keyhole shrinks from bottom to top as surface tension increases, allowing the liquid bridge to be merged. The molten metal shrinks and cools after one pulse period, creating two noticeable pits on Fig. 7 Evolution of temperature field and phase interface of case 3 in the first pulse period the upper and lower sides of the thin plate, which is quite different from the relatively flat weld obtained by PLBW without gap [10]. When the laser heat source is turned back on at 25.5 ms, the top of the liquid bridge substantially alters. As demonstrated in the right line of Fig. 10, it is clear that the right side of liquid bridge, which has solidified in the previous pulse cycle is remelted by the second pulse wave, resulting in another weld pool adjacent to the remained one. At 31 ms, the liquid bridge is separated again with the appearance of the keyhole, and a huge amount of liquid metal accumulates at the root surface of the workpiece. At 40 ms, the recoil pressure is soon reduced to zero as the laser powers off and keyhole shrinks again. The liquid bridge created in the second pulse period displays evident expansion in the welding direction as compared to the first pulse period.

Effects of welding speed and peak laser power on keyhole and weld pool dynamics
The created model was used to simulate PLBW under varying peak powers and welding speeds to investigate the influence of heat input on the dynamic behavior of the molten pool. Figure 11 depicts the keyhole and temperature field properties under various heat input circumstances. Due to the combined effect of surface tension and hydrostatic pressure opposite to recoil pressure, surface deformation occurs at the gas-liquid interface. The unevenness on the top surface and the bottom surface is a normal phenomenon to maintain the force balance on their surfaces. The dynamic evolution process of a molten pool simulated by diverse situations is comparable. The keyhole is always in three stages including that keyhole emerges first, then gradually creates a through-hole, the through-hole retains the oscillation condition, and ultimately the keyhole shrinks and vanishes when the laser power is reduced to zero. As presented at 2 ms in Fig. 11, when the laser peak power is increased from 2000 to 2400 W, the keyhole creation speed is accelerated, but the increase in welding speed has no significant influence. The keyhole profile becomes more curved as the laser peak power rises as shown at 10 ms, suggesting that the vibration of the keyhole wall is enhanced. The temperature field of the molten pool extends and the temperature of the keyhole wall rise as the laser peak power rises and the welding speed falls. As the welding heat input increases, the volume shrinkage of the molten pool becomes more visible due to the rise in the cooling time, as exhibited at 20 ms. For further research, Fig. 12 depicts the change in keyhole depth across two pulse cycles with varying peak powers. The keyhole depth varies regularly, keeping up with the three stages of keyhole evolution. The depth of the keyhole progressively grows, the maximum depth of the keyhole can reach 1.175 mm, then it remains horizontal, and ultimately, it gradually declines to zero. The vibration of the keyhole happens from time to time during the stage of keeping the depth of the keyhole horizontally. Using a peak power of 2000 W as an example, as the laser beam's propagation distance rises, the energy attenuation produced by Rayleigh scattering becomes more apparent, resulting in a fall in the recoil pressure at the bottom of the keyhole. At t = 13.96 ms, the keyhole vibrates fiercely, the depth of the keyhole drops to 0.875 mm, rebounds to 1.175 mm, then shrinks again at 15 ms, and disappears in the next microsecond. Comparing the keyhole depth under different peak power, it is discovered that the keyhole exists for the longest duration in the second cycle when the laser peak power is 2200 W. Figure 13 shows the evolution of the liquid bridge profile in a pulse cycle under different peak laser power conditions. The increased peak laser power makes it simple to build more liquid bridges on the gap at 1.5 ms when the laser is starting to melt the thin plate. With the increase of welding time, under the action of gravity, the liquid bridge under different peak laser powers shows different saggings. At 15 ms, the keyhole shrinks from bottom to top, and the separated liquid bridge generates a connection. When the peak laser power is varied, the liquid bridge profile changes somewhat. When the peak power is high, some of the liquid bridge is below the lower surface of the workpiece, and when it is low, the liquid bridge shrinks above the lower surface. This is because a higher peak power results in a larger welding heat input, longer cooling, and shrinkage times for molten metal. The liquid bridge under various peak laser power finished contracting at 25 ms. Figure 14 shows the evolution of liquid bridge morphology in a pulse cycle under different welding speeds. Comparing it can be found that there is no obvious difference in the liquid bridge profile at 1.5 ms, indicating that in the initial state of welding, the influence of welding speed changes on sheet melting is not obvious. When the welding speed increases, the heat input decreases, the melting amount of the sheet decreases, and the sagging phenomenon of the liquid bridge is improved, as shown in 5.5 ms.

Effects of welding speed and peak laser power on liquid bridge profiles
To further explore the influence of heat input conditions on the liquid bridge profile at the z = 0 plane, the liquid bridge area at different times in two pulse cycles was measured, as shown in Fig. 15. The liquid bridge area reflects the continuity Fig. 10 The shape of the liquid bridge at different times in the X-Y plane of z = 0 of the weld and the bridging ability of the molten metal. By observing Fig. 15a, it can be found that the peak laser power has a great influence on the liquid bridge area in the initial state, as previously discussed. With the increase of peak laser power, the liquid bridge area at different welding times increases synchronously. Under different peak laser power, the change of liquid bridge area with welding time tends to be consistent. Compared with the peak laser power, the welding speed has no significant effect on the liquid bridge area in the first pulse cycle, as shown in Fig. 15b. The area of the liquid bridge formed at different welding times grows in the second pulse period as the welding speed increases. And when the welding speed increases, the liquid bridge area expands more frequently, indicating that the welding speed increases and the molten pool's stability improves.

Conclusions
Using the developed multi-physical field CFD model, the dynamic simulation of the PLBW process of Ti6Al4V thin plate with 0.2 mm clearance was performed. At various initial welding stages, the evolution features of the temperature field, velocity field, and liquid bridge profile were examined. Using various welding heat input parameters, the transient temperature field, keyhole size, and liquid bridge area were computed. The effects of peak laser power and welding speed on the dynamic behavior of molten pools in the PLBW process were investigated. The following are the primary conclusions: (1) With a 0.2-mm clearance, the proposed PLBW technology and process parameters can weld thin plates of Ti6Al4V and produce a well-formed weld. The weld profile simulated by the CFD model is in good agreement with the experimental results, and the deviation is between 0.68 and 7.95%. (2) The molten pool's velocity field exhibits both molten metal convection and metal vapor eruption. While the flow rate of liquid metal is less than 2.0 m/s, the maximum eruption speed of metal vapor can reach 90.54 m/s. After the laser power reaches the peak value, the metal steam eruption weakens and the obvious Marangoni vortex appears in the molten pool. (3) The simulated keyhole is always in three stages-the keyhole appears and then transforms into a throughhole over time. The through hole continues to oscillate until the laser power is reduced to zero, at which point the keyhole ultimately shrinks and disappears. The keyhole shape becomes more curved as the laser peak power rises, suggesting that the keyhole oscillation is enhanced. (4) When the laser initially acts on the thin plate, the larger peak laser power tends to produce more liquid bridges in the clearance, while the influence of welding speed on the initial liquid bridge profile is not obvious. In the second pulse period, with the increase of welding speed, the area of the liquid bridge rises more regularly, indicating that the increase of welding speed enhances the stability of the molten pool.