Experimental study of the effect of the concentration of water/polyalkylene glycol solutions on heat transfer in steels subjected to quenching

In this work, the heat transfer coefficients were determined utilizing an experimental study for cooling a 15B35H boron steel, using aqueous solutions of polyalkylene glycol (PAG) as a cooling fluid. A cooling tank was designed and built to allow the fluid recirculation, where the effect of agitation and the PAG content in a concentration range of 2%–6% vol. were analyzed. The thermal histories of cylindrical probes instrumented with K-type thermocouples were obtained, and the heat transfer coefficients associated with the types of refrigeration were calculated, solving the IHCP. The results show the presence of two maximum surface heat flux denominated q1max for the low-temperature range and q2max for the high-temperature range. Getting higher q1max and low q2max is desirable to avoid distortion and cracking. It was found that 4% PAG exhibits a slight variation of heat flux values in both low and high-temperature regions, regardless of the degree of agitation, maintaining values of about 3 MW·m−2 for q1max and 5.5 MW·m−2 for q2max. In this case, q2max is higher than q1max, leading to higher cooling rates in the martensitic transformation zone, increasing the risk of distortion and cracking. In the cases of 2 y 6% PAG, agitation affects q1max ranging from 2.5 to 6 MW·m−2, and for q2max from 4 to 6 MW·m−2, where q1max is always higher than q2max, providing better conditions for quenching.


Introduction
In the heat treatments used in steels, quenching is employed to confer high hardness and resistance to the pieces, which can be treated in quenching media such as water, salt baths and oils.Its use leads to factors that could cause problems in the final properties of the material, such as unwanted hardness profiles, microstructural distortion and cracking.Most cases are due to the cooling mechanisms in standard vaporizable quench media such as water and oils.Those quenchants provide a nonuniform heat extraction due to the presence of a vapor film and its evolution during the cooling process, causing the soft spots phenomena and leading to microstructural distortion.The effect of the different parameters during the quenching is fundamental and thus elucidates the optimal conditions to obtain the specific properties in steel.Polymeric polyalkylene glycol (PAG) solutions arise as an alternative to these problems.Due to the inverse solubility properties of PAG and glycol polymers, these quenchants minimize residual stress and distortion in the quenched parts by offering more uniform heat transfer than water.The polymer separates from water when the temperature of the PAG-water solution is raised above the cloud point temperature.The vapor film on the hot metal surface is eliminated through the encapsulation by a uniform polymer film layer.The polymer film rupture is almost instantaneous and increases the uniformity of heat transfer [1].Previous investigations have been carried out using glycol polymers and other types of polymeric solutions in quenching; Chen et al [2], conducted a comparative study of different quenching media in applications such as glycerol, water, oil and PAG.Their results show that a 5% glycerol solution has superior quenching performance, including better case hardening compared to PAG and higher crack resistance than water.The last is attributed to a high coefficient of heat extraction by convection (h) in the high temperature stage, while it exhibits a low h during the convective cooling stage.Kobasko [3], studied reducing distortion in tempering processes with PAG as a cooling medium.He shows that at the end of cooling, the polymer can be locally dissolved by a cold water flow, creating a local open area where the martensitic transformation begins.Due to the higher percentage of martensite, a large distortion is created, so he proposes to interrupt the cooling process or stop the stirring before the insulating coating dissolves.Also, Kobasko [4] mentions that the use of low-concentration polymer solutions favors the creation of a thin polymer insulating layer during tempering, decreasing the initial heat flux density, which is why the boiling process is not carried out, which makes it possible to use steel of optimum hardenability instead of alloy steels containing expensive alloying elements.Mathews et al [5], performed quenching experiments on different steels and cooling media.They observed a more even hardness distribution using distilled carbonated water compared to other quench media.The propensity for quenching defects in steel parts hardened in carbonated water, water, and PAG media diminished with decreasing martensite start temperature (M s ).The high quench rates obtained by water and carbonated water below M s of steels often resulted in cracks and higher residual stress in high and medium carbon steel parts.Kobasko [6], experimentally determined the behavior on cooling of a 1% aqueous solution of PK-2 polymer, which is a solution of iron-containing salts and polyacrylic acid with additives of neutralizing component.He found that the surface heat flux density curves present two maximum heat flux densities, caused by a first and a second boiling crisis.Ikkene et al [7], investigated the influence of polymer concentration, temperature and agitation on the cooling performance of water-based polymer solution poly(ethyloxazoline) (PEOX).They found that contents of 2.5 and 5% show a rapid transition from vapor to boiling phase, 10 and 15% show a gradual transition due to the formation of the polymeric layer.Increasing temperature and polymer concentration reduce the cooling rate, while increasing agitation decreases the stability of the vapor layer and increases the cooling rate.Momoh et al [8], used polyethylene glycol (PEG) solutions in different proportions as quenching medium of a low-alloy medium carbon steel.They evaluated the effect on mechanical properties and microstructural evolution.The hardness decreases with increasing PEG concentration, while impact energy reveals a behavior contrary to tensile strength.However, the ductility diminished with decreasing percentage of the polymer.The above is caused by the high proportion of martensitic phase obtained.Ramesh and Prabhu [9] conducted studies on the effect of glycol polymer concentration on the cooling performance of Inconel 600 specimens.They found that low concentrations provide more uniform cooling due to rapid rewetting.At the same time, high glycol contents polymer exhibited non-uniform slow and repeated rewetting, causing a large variation in heat transfer over the surface of the specimens.Banerjee and Waghmare [10] quenched a 41Cr4 steel, using PEG as a cooling medium in concentrations between 5%-9%.Their results determined that the use of polymeric solutions showed a positive effect, obtaining the desired hardness and reduction of cracking.Zainulabdeen et al [11] used a polyvinylpyrrolidone (PVP) polymer solution in concentrations of 15%-25% and different bath temperatures in medium carbon steel quench.The results showed that the best combination of properties was achieved with samples quenched at bath temperature of 40°C with 20% PVP, determining that this treatment can be used instead of the quench with water for medium carbon steel samples.This work investigated the heat transfer in quenching by immersion in boron steel parts using PAG aqueous solutions.The effect of PAG content was determined in a range of low concentrations (2, 4 and 6 %vol.), as well as the effect of agitation by jets using flows of 3, 6, 9 and 12 l•min −1 in a quenching tank.It was found that small changes in PAG concentration could considerably affect surface heat flux density.The effect of the PAG content in a range of low concentrations (2, 4, and 6% vol.) and the agitation employing jets using flows of 3, 6, 9, and 12•min −1 on the behavior of heat transfer curves were analyzed.

Experimental methodology 2.1. Experimental setup
Figure 1 shows the experimental system where the quench tests were carried out.The quenching setup consists mainly of a vessel with jet grids that provide cooling of the fluid on the hot surface of the samples.In figure 2(a), the dimensions of the quench vessel are shown, and an enlargement of the jet grid and its dimensions can also be seen.Each jet is connected to a main supply line, which in turn is equipped with a high-pressure pump that allows the regulation of the flow rate required for cooling with the help of a pressure gauge.

Measurement of thermal records
For the quenching tests, cylindrical AISI-SAE 15B35H steel probes were machined with dimensions such as those shown in figure 2 b); these were instrumented with a K-type thermocouple introduced to a depth of 30 mm and at a distance 5 mm from the radial surface.The recording of temperatures was carried out using a FLUKE NetDAQ 2645A data acquisition system with a sampling frequency of 10 Hz.The specimens were  quenched in Water/PAG polymer solutions in different concentrations of 2, 4, and 6% vol. of PAG Aquaquench 245, manufactured by Houghton International Inc.To obtain the thermal histories, the average of five tests performed for each quenchant was taken.The procedure consisted of heating the probes in a vertical resistance furnace until reaching the austenitizing temperature of 870 °C (1143K), with a permanence period of 15 min.Next, the pumping system was turned on until a specific fluid level was reached in the quenching vessel, at which time flow rates were regulated with the help of a valve.Afterward, the probes were subjected to quenching in the polymer solutions at a bath temperature of 25 °C.The different flow conditions and cases studied are shown in table 1.To guarantee the reproducibility of the experiments, five cylindrical 304L steel specimens with dimensions of 60 mm in length and 10 mm in diameter were subjected to quenching by immersion without stirring according to the D-6482 standard in different cooling media such as water, oil, and polymer solution 6% PAG.

Solution of the inverse heat conduction problem
Once the thermal records were obtained through the experimental system, the calculation of the heat transfer through the jets was determined by solving the IHCP.The Energy equation (1) was solved iteratively where r (m) is the radial coordinate of the specimen, ) are the thermal conductivity, density, and specific heat of the material, respectively.The details of the IHCP solution method of equation (1) are previously described [12][13][14].
For the solution of equation (1), the following boundary conditions were used: on the symmetry axis of the specimen there is no heat transfer, so an isolated boundary is considered as shown in equation (2 The heat that arrives by conduction at the surface in contact with the fluid is transferred to the surroundings by radiation and forced convection represented in equation (3) by a global heat transfer coefficient h.
The surface heat flux density is estimated from knowledge of the measured temperature inside a heat conducting solid by minimizing the function in the equation (4) where M = θ • t −1 , at different finite regular intervals.T n and Y n are the calculated temperatures measured at a location near the metal/fluid interface, respectively, θ and t are the time steps for heat flux and temperature estimation, respectively.Applying the minimization condition, the heat flux correction is estimated at each iteration step.This procedure is continued until a convergence criterion of less than 0.005 is obtained.This procedure allows obtaining simultaneously the surface temperature of the sample in contact with the cooling medium and the interfacial heat flux.Subsequently, by means of equation (5) of Newton's Cooling Law, where q is the heat flux at the solid/liquid interface, T is the calculated surface temperature and T f is the fluid saturation temperature, the heat transfer coefficients h of the surface in contact with the quenching fluid are calculated.

= -
h q T T 5

Experiments reproducibility
In order to obtain thermal records, strict control of all the process variables is required.In the experimental setup developed, one of the most critical variables is the agitation inside the cooling vessel, which is controlled until it reaches a stable flow.Figure 3(a) shows the reproducibility of a series of five experiments carried out with different cooling media without stirring in 304L steel specimens, where the standard deviation is shown to guarantee the replication of the experiments.These experiments were carried out with 304L steel served, on the one hand, to check the reliability of the experimental system as well as to show the typical behavior of the cooling media by discarding the effect of oxide formation on the surface as occurs in the heat treatment of conventional steels.It is observed that the polymer solution shows a more significant variability of the data compared to the tests carried out in water and oil.These variations are attributed to the behavior of polymer solutions with inverse solubility and the fluid's wettability to the piece.According to Chandler [15], increasing the polymer concentration increases the thickness of the film formed by the polymer, and the wettability of the surfaces of the parts improves with 5 %PAG solutions.Additionally, figure 3(b) shows the cooling rate curves of the different quenching media.It can be seen that water has the highest cooling rate, followed by the polymer solution and oil.The cooling process for the case of oil and polymer solution shows the three stages of cooling highlighted with marks, vapor layer, nucleated boiling, and convective cooling.However, in polymeric solutions, according to Tensi [16], the heat transfer mechanisms differ, since when a probe is quenched, its temperature is sufficient to reach the inverse solubility point of PAG, which gives rise to the formation of a stable polymer-enriched vapor film that encapsulates the surface, promoting slow cooling, analogous to the vapor phase in oil cooling.As the temperature of the hot surface decreases, the vapor layer breaks up explosively, up to approximately the rewetting temperature, where the fluid comes into contact with the hot metal surface, resulting in a boiling process pseudonucleate with high heat extraction rates.The boiling period advances, and cooling occurs by conduction and convection in the liquid.When the temperature drops below the inversion temperature (60 °C-90 °C), the polymer re-dissolves into a homogeneous solution, at which point a second maximum cooling rate occurs.In the case of water, stable vapor layer formation does not occur because cooling is a progressive "shock film" boiling sequence where bubble nucleation first occurs about 0.1 seconds after the hot metal is first immersed in the liquid.In this stage, the vapor bubbles first move away from the hot metal surface after their formation, rapidly reaching the critical heat flux.

Effect of PAG concentration and agitation on heat transfer
figure 4 shows the cooling velocity curves of the tests on 15B35H steel in different polymer solutions with stirring.It can be seen that the cooling stages are seriously affected by the agitation, revealing the non-existence of the formation of a stable layer of steam.Figure 5 shows the evolution of the detachment of the oxide layer formed during heating, which occurs during the first fractions of a second of the contact of the hot piece with the cooling medium, which, with the help of the impact of the jets of fluid on the surface, it quickly detaches, leaving an irregular surface, showing areas of little contact with the fluid and causing fluctuations in the curves during heat extraction.According to Barrena et al [17], the wall surface roughness affects single-phase convection since it improves the specific surface area, increasing the heat flux.No oxide layer remnants were found after the treatments.On the other hand, the formation of the polymer-rich layer is not formed as it usually does, but rather the formation of a higher density area around the piece is favored without reaching the point of the insolubility of the polymer, which over time it completely dissolves as the cooler recirculating liquid combines in this zone.Once the polymer is dissolved, a second heat extraction peak occurs in a temperature range of 100 °C-  400 °C.Table 2 shows the compilation of the characteristic values of the cooling velocities curves of the tests carried out on 15B35H steel, such as the two maximum points of heat extraction CR1 max and CR2 max , at the temperatures and times in which these occur T1, T2, t1 and t2 respectively.The CR 300 °C is also shown, corresponding to the cooling rate in the martensitic transformation range and the times elapsed at temperatures 600, 400 and 200 °C.As the PAG content increases, the CR1 max parameter decreases compared to that of water since the addition of PAG in water decreases the severity of the quench.However, the CR1 max and CR2 max for the polymeric solutions are higher in the case of the addition of 4 %vol. of PAG.On the other hand, while comparing 2, 4 and 6%vol. of PAG, the maximum of the CR1 max parameter occurs at different agitation rates, 12, 3 and 6 l•min −1 respectively.In the parameters t 600 °C, t 400 °C and t 200 °C, it is observed that the polymer solution exhibits longer cooling times as the PAG content increases; however, a 4 %PAG content, shows shorter times regardless of the agitation which is evidenced in the t 200 °C parameter.
In figure 6 the surface heat flux is shown as a result of solving the inverse heat conduction problem.The fluctuations in the heat flux can be distinguished in addition to the presence of two maximum heat fluxes during cooling, the first q1 max within a temperature range between 600 °C-800 °C and the second q2 max between 100 °C-400 °C, the latter being of greater magnitude in most cases.These fluctuations indicate that the heat extraction is interrupted in the first instants of cooling due to the instability of the formation of the vapor layer mentioned above.Once a temperature close to the re-dissolution temperature of the polymer is reached, and when the liquid close to the piece is mixed with colder liquid, the characteristic explosion of polymer solutions occurs, giving way to the convective heat extraction mechanism, presenting an increase in heat flux.According to [17,18], the uniformity of the heat flow is critical to minimize thermal and phase transformation stresses, helping to reduce distortion and cracking problems.From the above, it can be mentioned that a more negligible difference between the maximum heat flux q1 max and q2 max reduces the risk of distortion and cracking of the material.For a content of 2% PAG, q1 max tends to decrease as the degree of agitation increases from values between 4-5 MW•m −2 to 3 MW•m −2 , while q2 max shows magnitudes of between 3.8-5.5MW•m −2 .For a 4% PAG content, q1 max is not significantly influenced as the agitation increases 6-12 l•min −1 with low values of approximately 3 MW•m −2 , maintaining flows of this magnitude until reaching the beginning of the convection stage presenting lower fluctuations in the heat flux, unlike the other cases, while q2 max rises to magnitudes between 5-6 MW•m −2 .From the previous cases, it can be seen that for agitation of 3 l•min −1 higher values of q1 max occur; for a content of 2% PAG, both q1 max and q2 max remain in a range of 4-5 MW•m −2 , while for a content of 4% PAG, q1 max has a magnitude of approximately 6.5 MW•m −2 .At the same time, as cooling takes place, it decreases to values of up to 3.7 MW•m −2 in the stage transition to subsequently increase q2 max to values of 5.8 MW•m −2 .A content of 6% PAG shows similar behavior to 2% PAG.However, it shows a tendency to decrease q1 max as the degree of agitation increases.It should be noted that, for this last PAG content, the agitation used by a flow of 3 l•min −1 does not suddenly increase the value of q1 max as in the case of 2 and 4% PAG. Figure 7 shows the curves of the heat transfer coefficients calculated from the heat fluxes and surface temperatures.It can be observed that a content of 6% PAG shows h fluctuations in the first moments of cooling, which indicates an increase in the heat extraction rate corresponding to the mechanism of heat extraction by boiling.In contrast, the contents of 2 and 4% PAG do not exhibit this increase in the coefficient during the boiling stage in the cooling, except for agitation of 3 l•min −1 , where higher heat extraction rates are observed.However, a content of 2% PAG shows the onset of the convective stage at an earlier stage at a temperature of about 280 °C compared to a content of 4% PAG, in which the convective stage is delayed until temperatures below 200 °C.It is to be expected that as the PAG content increases, the heat transfer coefficient will decrease; however, in this study, it has been found that a 4% PAG content shows a greater homogeneity of heat extraction over time.It exhibits lower heat extraction rates throughout the cooling process than in 2 and 6% PAG contents.

Effect of PAG concentration and agitation on hardness of 15B35H steel
Although the difference between the PAG contents of the quenching media is slight, it was found that there are significant differences in the quenching rates provided by the different quenchants.Figure 8 shows the CCT diagram calculated using JMatPro software, in which the lines of the maximum and minimum cooling rates achieved by the quenchants are found.It shows that the 2% PAG quenchant with an agitation of 6 l•min −1 provides the maximum cooling rate reached, 100 °C•s −1 , while the minimum is for the 6% PAG quenchant with agitation of 6 l•min −1 with 70 °C•s −1 .The diagram suggests that the microstructures present in the different steel parts are mainly martensitic for the higher cooling rates.As cooling rates decrease, the bainitic structure begins to appear, causing the decrease of hardness.In figure 9, the Rockwell hardness of the specimens is plotted as a function of PAG content and the degree of agitation.It shows a tendency for 2 and 4% PAG to enhance the hardness with the agitation of 6 l•min −1 concerning 3 l•min −1 ; however, when the agitation is more vigorous, the hardness is decreased.The opposite happens for the case 6% PAG, which presents a decrease in hardness with agitation of 6 l•min −1 concerning 3 l•min −1 , while by increasing the degree of agitation to 9 and 12 l•min −1 , an increase in hardness is observed.This is consistent with the fact that higher cooling rates provide higher hardness values; however, it is to be expected that as the PAG content increases, the hardness decreases; however, it was found that a 6% PAG content can provide higher hardnesses values if used with a high degree of agitation.

Conclusions
With the experimental system employed, it was possible to satisfactorily obtain the heat flow curves for different quenching media in an agitated vessel employing jets, varying the flows and the PAG concentration where the following can be concluded: • The constant fluctuation of q during cooling between the parameters q1 max and q2 max is attributed to the fact that agitation interrupts the formation of a stable vapor layer, favoring the coexistence of the vapor layer and boiling cooling mechanisms.
• Surface heat flux density is sensitive to small changes in PAG concentration.• A 4% PAG concentration shows similar values of q1 max and q2 max , regardless of the degree of agitation, but higher cooling rates in the martensitic transformation zone, increasing the risk of distortion and cracking problems, except for the agitation of 3 l•min −1 .
• In the 4% PAG solution, q2 max is higher than q1 max ; this causes a significant susceptibility of the treated parts to present distortion and cracking problems.
• In the 2 and 4% PAG solutions, an agitation provided by 3 l•min −1 provides higher hardness; the opposite occurs with 6% PAG; in the same agitation degree, the lower hardness is obtained.
• In using polyalkylene glycol aqueous solutions as a cooling medium, it is difficult to obtain good reproducibility in the cooling rates since the phenomena that occur due to the formation of the polymer layer in the surroundings of hot parts are unpredictable.

Figure 2 .
Figure 2. (a) Design of the fluid jet outlets; (b) Dimensions of the fabricated specimens and location of the measurement point (K).

Figure 3 .
Figure 3. (a) Experimental reproducibility of measured thermal histories, with corresponding standard deviation; (b) Cooling rate curves for immersion quenching tests on 304L steel using water, oil and 6 %PAG aqueous solution as quenchants.

Figure 4 .
Figure 4. Cooling rate curves of AISI-SAE 15B35H steel quenched by immersion in aqueous polymer solutions at different PAG concentrations and agitations.

Figure 5 .
Figure 5. Detachment of oxide layer from the surface during quenching of AISI-SAE 15B35H steel quenched by immersion in 2% PAG polymer solution, with agitation of 3 l•min −1 .

Figure 6 .
Figure 6.Heat flux at the solid/liquid interface of AISI-SAE 15B35H steel quenched in PAG aqueous solutions at different concentrations and stirring.

Figure 7 .
Figure 7. Convective heat transfer coefficients of AISI-SAE 15B35H steel quenched PAG aqueous solutions at different concentrations and stirring.

Figure 8 .
Figure 8. Calculated CCT diagram for 15B35H steel; the superposed lines show the maximum and minimum cooling velocities for the quenchants.

Table 2 .
Characteristic values of cooling rate curves for aqueous PAG polymer solutions on 15B35H steel at different concentrations and stirring speeds.