THERMAL CYCLING EFFECTS ON PROPERTIES OF STEEL GRADE USED FOR HEAT EXCHANGERS: DOE PPROACH

Ashraf M. Elaieb1, F. M. Shuaeib1, Ramadan O. Saied2 and Mohammed A. Gebril1

1

Mechanical Engineering Department, Faculty of Engineering Garyounis University, Benghazi, Libya 2 Mechanical Engineering Department, Faculty of Engineering, Al-Fateh University, Tripoli, Libya

E-mail: drfmshuaeib@yahoo.com, Tel: 00218-926225683

Abstract - In this paper, the influence of thermal cycling on the corrosion and hardness has been studied for low carbon steel material. This type of steel is cheap and extensively used in industry including heat exchangers, cooling towers and similar heat transfer equipments. During the normal operation of these equipments, this material is subjected to thermal cycles (heating and subsequent cooling) which are usually below the first phase transformation temperature for steel ( 723oC). Previous work was either considering standard heat treatment methods or using the classical method of changing one parameter each time. In this work the temperature range selected is below the 723 oC and the response surface methodology in the Design of Experiment (DOE) was implemented to tackle these shortcomings. Therefore low carbon steel specimens were prepared and thermally cycled for 20, 30 and 40 cycles at temperatures of 400, 500, and 600 oC respectively. During one thermal cycle the specimen underwent heating and rapid cooling in water. Corrosion and hardness were measured for the selected number of cycles and thermal cycling temperatures. The influence of thermal cycling on the hardness and corrosion of the low carbon steel was determined and analyzed and interesting conclusions were drawn.

many applications , including structural beams, car parts and bodies, kitchen appliances, cans, pipe line, railways, tractors and agriculture implement, petrochemical and engineering industries, the list is endless [1 and 5].

In like these applications the carbon steel parts may be subjected to heating then cooling many times throughout the operation; these heating and cooling is called thermal cycling. When a hot part is rapidly cooled in a solution such as water that produces a high heat transfer, the temperature differences create high thermal stresses which often cause deformation [56]. This distortion works on scattering of atoms and increasing the distortions in the microstructure and also works on the creating of stresses. These internal reasons resulting from the operations of the heating and cooling and changes in its dimensions (expansion and contraction), results on an increase in hardness, and from nature of the metal if it is increased in hardness, would be more resistant to wear (particularly erosion), but these stresses and these distortions may make the metal more susceptible to corrosion on the contrary than it is Keywords - Thermal cycling, corrosion, heat exchanger, hardness, in the hardness and wear resistance [7]. This may affect the low carbon steel mechanical and tribological properties of the metal. This phenomenon is observed in many of applications as in various I. INTRODUCTION types of heat transfer equipments such as heat exchangers and This Steel is the worlds most important material, multi- boilers (See Fig. 1). functional and most adaptable material. About 5% of iron element is present in earth's crust [1]. Without steel, the world as we know it would not exist: from oil tankers to thumb tacks, from trucks to tin cans, from transmission towers to toasters [2]. Or we can say steel is arguably worlds most "advanced material. It is very versatile material with a wide range of attractive properties which can be produced at a very Fig.1 Corrosion and erosion of heat exchangers tubes competitive production cost [3]. Carbon steel is by far the most widely used kind of steel. The properties of carbon steel depend primarily on the amount of carbon it contains. Most carbon steel has a carbon content of less than 1% [4]. Low carbon steel is easily available and cheap having all material properties that are acceptable for Jokhio et al. [7] conducted various types of heat treatment on carbon steel and found that water quenching of low carbon steel specimen has the highest tensile strength and hardness. However this treatment gives the lowest ductility compared to other treatments. They concluded that quenching is

recommended when the strength and hardness is the prime factor in design. The microstructure of steel consists of matrix of ferrite and pearlite. The grain fines of pearlite increases by increasing the rate of cooling. However, they have not considered the temperatures below the first transformation temperature (A1). T. Foley and A. Levy [8] conducted erosion study on low carbon steel and found that the ductility of the steels (and consequently hardness) had a significant effect on their erosion resistance which increased with decreasing ductility. This suggests that the hardness can be utilized as an erosion resistance performance parameter. II. MATERIALS AND METHODS A. Material Table (1) shows the chemical composition of the low carbon steel used in this investigation. The material which has ASTM code A576 Grade 1018 has very close chemical and mechanical characteristics to the ASTM A214 heat exchanger and condenser tube material and is chosen because it is available on rods which is suitable for the research work requirements [8]. Specimens were cut from a rod of 20 mm diameter in the form of cylindrical shape with 20mm in diameter and 10mm in length (See Fig.2). The specimens, in the as-received condition, were annealed to get rid of any stresses which might be introduced during manufacturing and cutting. Table (1): ASTM-A576, Grade1018 Chemical composition Element C Mn ph S % 0.14 - .21 0.25 -0.4 0.4 0.05

to homogenize at that temperature for 1 hour. After that the specimens were immediately (rapidly) cooled in water at a constant ambient temperature (between 25 to 35 oC). This constitute one cycle. For each experiment No. the previous steps were repeated to the required number of cycles. After all experiments were completed, specimens were then taken out for hardness, and corrosion testing. The hardness values were determined using 100kg load on Rockwell (HRB- kg/mm2) hardness tester. The corrosion test was based on subjecting the specimens to the environmental (atmospheric) corrosion for a period of 3 months. This period was decided after few trials for determining the suitable exposure period that would produce sufficient amount of corrosion that can be measured. III. RESULTS AND ANALYSIS After the RSM matrix is completed by introducing the results as shown in Table (1), the design matrix was analyzed using RSM routines in MINITAB software. The full quadratic model was selected to analysis the response surfaces. The output results were generated in the form of contour plots and three dimensional response surface plots; statistical models were also produced.

Fig. 3 The thermal cycling process

Specimen dimensions Fig. 2: test specimen

Physical shape

TABLE 1 THE RSM DESIGN MATRIX

B. Methods of Experiments In this work a response surface methodology (RSM) design of experiment was implemented to plan the experimental work. The RSM range was selected based on literature values and also after performing few trials. After the design factors and their levels were set, the design of experiment matrix based on the Central Composite RSM method was created using MINITAB software (See Table 1). Subsequently, the thermal cycling processes, the hardness, and corrosion performance tests were carried out according to the designed experiment matrix listed in Table (1). C. Description of the Thermal Cycling Process This process involved the steps as shown schematically in Fig. (3). Specimens for each temperature and number of cycles setting (Exp. No) were heated to the required temperature according to Tables (1). Then they were allowed

Exp. No 1 2 3 4 5 6 7 8 9

Number of cycles 30 20 20 40 40 40 30 30 20

Cycling Temp. oC 600 400 500 400 500 600 400 500 600

Hardness (HRB) 81 81.5 85.4 83.1 91 80.5 82.6 88.4 83.3

Corrosion Wt loss ( g) 0.58 0.498 0.373 0.449 0.361 0.982 0.513 0.339 0.653

A. Hardness RSM Analysis The hardness results were analysed using response surface methodology (RSM), and the interactions of thermal cycling parameters were identified. 1) The RSM model: Regarding the statistical model, the statistical analysis for the full quadratic model hardness results resulted in coefficients of determination of R2 and R2-Adj = 86.7% and 65 % respectively, which shows that the performed hardness regression is very good (see appendix-A). The RSM model showed small p-values for the thermal cycling temperature and thermal cycling temperature squared (p = 0.023 and p = 0.025 respectively) suggesting these effects are important, slightly large p-value for the interaction of cycling temperature and number of cycles (p = 0.374 ) and intermediate p-value for the number of cycles (p = 0.639); this suggesting that the main effecting processing parameters were the thermal cycling temperature, and the less effecting one was the number of cycles, while the interaction of cycling temperature and number of cycles had an intermediate effect. Therefore, the verified model of hardness is as follows: HRB = - 83.989 + 0.54334 C + 0.65567 T+ 0.00134 C2 -0.00063T2 - 0.00110 CT . ..(1) Noting that, HRB is the hardness in RB, C is number of cycles, T is cycling temperature (oC). 2) Hardness RSM parametric study: The effects of thermal cycling processing parameters on hardness are presented in Figures (4) and (5) as 3D surface plot and contour plot obtained using MINITAB regression model of equation (1). All these graphs are held and identified at middle values of processing parameters. As shown in the 3D- surface plot of Figure (4), the effect of the cycling temperature is more significant than the number of cycles. This may show that either the number of cycles is not that much significant as compared to the thermal cycling temperature or that the selected cycles range is not sufficient to clearly explore its effects. This is recommended to be as a future work in this subject. Regarding the number of cycles, the increase is almost linear within the selected range. However, the effect of the cycling temperature is increasing the hardness in a nonlinear curve until a peak at about the 500 oC, and then it is decreasing again. The overall response looks like a tilted saddle. These findings are confirmed by the contour plot of Figure (5), which shows that the effect of the number of cycles may reasonably be represented by horizontal lines. The effect of the cycling temperature is very clear with a region of maximum hardness around the 500 oC temperature. This temperature is the start of the grain re-crystallization where after this temperature stress reliving would usually take place. Before this 500 oC temperature hardness increases due to residual stresses building up. This residual stresses are generated from the rapid cooling of the thermal cycling

processes. However, for the 600 oC temperature results this residual stresses would be relieved each time the specimens are reheated for the subsequent cycle.

Fig.4 The 3D-surface plot of the hardness as a response and cycling temperature and No. of cycles.

Fig. 5 The contour plot of the hardness as a response and cycling temperature and No. of cycles.

B. Corrosion Rate RSM Analysis. The corrosion rate results were analysed using response surface methodology (RSM), and the interactions of thermal cycling parameters were identified. 1) The Corrosion rate RSM model: Regarding the statistical model, the statistical analysis for the full quadratic model for corrosion rate measurements resulted in coefficients of determination of R2 and R2-Adj = 88.9 % and 70%, which shows that the performed corrosion rate regression is good (see appendix-A). The RSM model showed small p-values for the thermal cycling temperature and thermal cycling temperature squared (p = 0.041and p = 0.045 respectively) suggesting these effects are important, slightly large p-value for the number of cycles and the interaction of cycling temperature and number of cycles (p = 0.2 and p = 0.18 respectively) and intermediate p-values for the number of cycles squared (p = 0.40); this suggesting that the main effecting processing parameters were the thermal cycling temperature, and less effecting one was the number of cycles and the interaction of cycling temperature and number of

cycles, while the number of cycles squared had an intermediate effect. Therefore, the verified model of corrosion rate is as follows: CR (g) = 8.0106 - 0.08799 C -0.02706 T + 0.00075 C2 + 0.00003 T2 + 0.00009 CT . ..(2) Noting that, CR is the corrosion rate expressed as the weight loss in grams, C is number of cycles, T is cycling temperature (oC). 2) Corrosion rate RSM parametric study: The effects of thermal cycling processing parameters on corrosion rate are presented in Figure (6) to Figure (7) as 3D surface plot and contour plot obtained using the MINITAB developed model of equation (2). As shown in the 3D- surface plot of Figure (6), the effect of the cycling temperature is more significant than the number of cycles. However, based on the results obtained, regarding the number of cycles, the decrease of corrosion rate with increasing the number of cycles is almost linear within the selected range. The effect of the cycling temperature is decreasing the corrosion rate in a non-linear curve until a minimum at about 500 oC, and then it is increasing again. These findings are confirmed by the contour plot of Figure (7), which shows the effect the number of cycles as approximately horizontal lines. The effect of the cycling temperature is very clear with a region of minimum corrosion rate around the 500 o C temperature. This temperature is the start of the grain recrystallization where after this temperature stress reliving would usually take place. Before this 500 oC temperature, corrosion was high and decreases due to residual stresses building up (more stored energy). This residual stresses are generated from the rapid cooling of the thermal cycling processes. However, for the 600 oC temperature results even though this residual stresses would be relieved each time the specimens are reheated for the subsequent cycle, the corrosion rate increases again due to oxidation which is reported to start at about 570 oC.

IV. CONCLUSIONS From this experimental study the thermal cycling temperature was found to have more significant effects than the number of cycles on both the hardness and the corrosion rate. Increasing the thermal cycling temperature generally was found to increase the hardness until about 500oC where peak hardness is obtained, and then it is followed by a decrease in the hardness.

Fig.7 The contour plot of the corrosion rate as a response and cycling temperature and No. of cycles

Increasing the thermal cycling temperature also, decreased the corrosion rate until it comes to a minimum at about 500 oC then it increases again. However, increasing the number of cycles, was found to increase the hardness but not to the same extent of the temperature and increase the corrosion rate but not to the same extent of the temperature. It is interesting to note that there is a temperature where peak response values were obtained for both hardness and corrosion tests which is the 500oC. This seems to be a critical temperature in this steel type and designers need to be aware of the changes occurring around it. This type of material showed weak hardness and corrosion properties as the temperature goes above 500 oC, which suggests that this material may not be suitable above this temperature for applications where the performances considered are essential requirements. Finally, empirical models were developed which predicts the hardness and the corrosion rate performance of this type of steel within the selected temperature and number of cycles range. This is a good achievement as no model could be obtained from the previous literature on these issues. REFERENCES
[1] [2] [3] www.sail.com accessed at 2010 www.Iron and Steel Industry .com/ The Canadian Encyclopedia.htm . Accessed at 2010. Bello K.A., Hassan S.B. and Abdulwahab M. Effects of Tempering on the Microstructure and Mechanical Properties of Low Carbon, Low Alloy Martensitic Steel, Department of Metallurgical Engineering, Ahmadu Bello University Zaria, Nigeria. Journal of Applied Sciences Research , 2007, page 2. Sanjib kumar jaypuria.," A Project Report on Heat Treatment of Low Carbon Steel", Department of Mechanical Engineering National Institute of Technology, page 8, Session 2008-09, Under the guidance of Prof. (Dr.) S. K. Patel. The ASM Committee on, "Alloy phase diagram ", ASM Metals Handbook, page 232 , vol 3, 2005.

Fig. 6 The 3D-surface plot of the corrosion rate as a response and cycling temperature and No. of cycles.

[4]

[5]

[6]

M. H. Jokhio, S. A. Memon, and M. J. Sangi, "Effect of heat treatment on the microstructure and mechanical properties of low carbon ASE1010 steel", E-Awam University research journal of engineering science & technology, vol 1, no 1, page 6., 2000.

[7] [8]

T. Foley and A. Levy., "The erosion of heat-treated steels", Wear, Volume 91, Issue 1, pp. 45-64, (1983). Annual Book of ASTM, Standards, ASTM A576 Grade 1018 & ASTM A214 heat exchanger and condenser tube material.

APPENDIX-A: RSM REPORTS A1-Response Surface Regression: Hardness versus Cycles and Temperature The analysis was done using un-coded units.
TABLE A-1 ESTIMATED REGRESSION COEFFICIENTS FOR HARDNESS MODEL

Term Constant Cycles Temperat Cycles*Cycles Temperat*Temperat Cycles*Temperat S = 2.114

Coef -83.99 0.54 0.66 0.00 -0.00 -0.00

SE Coef 41.8941 1.0445 0.1530 0.0149 0.0001 0.0011

T P-value -2.005 0.139 0.520 0.639 4.284 0.023 0.089 0.935 -4.193 0.025 -1.041 0.374

R-Sq = 86.7%

R-Sq(adj) = 64.6%
TABLE A-2 ANALYSIS OF VARIANCE FOR HARDNESS

Source Regression Linear Square Interaction Residual Error Total

DF 5 2 2 1 3 8

Seq SS 87.604 4.187 78.578 4.840 13.404 101.009

Adj SS 87.6044 82.0385 78.5778 4.8400 13.4044

Adj MS 17.5209 41.0192 39.2889 4.8400 4.4681

F 3.92 9.18 8.79 1.08

P-value 0.145 0.053 0.056 0.374

A.2. Response Surface Regression: Corrosion rate versus Cycles and Temperature The analysis was done using un-coded units.
TABLE A-3 ESTIMATED REGRESSION COEFFICIENTS FOR CORROSION RATE MODEL

Term Constant Cycles Temperat Cycles*Cycles Temperat*Temperat Cycles*Temperat S = 0.1088

Coef 8.01061 -0.08798 -0.02706 0.00075 0.00003 0.00009

SE Coef 2.15652 0.05377 0.00788 0.00077 0.00001 0.00005

T 3.715 -1.636 -3.435 0.979 3.312 1.737

P-value 0.034 0.200 0.041 0.400 0.045 0.181

R-Sq = 88.9%

R-Sq(adj) = 70.4%
TABLE A-4 ANALYSIS OF VARIANCE FOR CORROSION RATE

Source Regression Linear Square Interaction Residual Error Total

DF 5 2 2 1 3 8

Seq SS 0.283926 0.106975 0.141230 0.035721 0.035518 0.319444

Adj SS 0.283926 0.159198 0.141230 0.035721 0.035518

Adj MS 0.056785 0.079599 0.070615 0.035721 0.011839

F 4.80 6.72 5.96 3.02

P-value 0.114 0.078 0.090 0.181