INFLUENCE OF HEAT TRANSFER DYNAMICS ON HARDNESS DISTRIBUTION AFTER QUENCHING
VPLIV DINAMIKE PRENOSA TOPLOTE NA PORAZDELITEV TRDOTE PO KALJENJU
BOŽIDAR LIŠČIC
Faculty of Mechanical Engineering and Naval Architecture, Ivana Lučiča 5, Zagreb, Croatia
Prejem rokopisa - received: 1997-10-01; sprejem za objavo - accepted for publication: 1997-10-21
The pattern of hardness distribution on round bars1 cross-section after quenching was studied in relation to the change of heat transfer on the workpiece surface. It was found that a 'delayed quenching', producing a discontinuous change of cooling rate, may result in higher hardness in the core, than at the surface. This phenomenon called 'inverse hardening' has been theoretically explained by Shimizu and Tamura. It depends on: hardenability of the steel, cross-section size of the workpiece and on quenching condition, and is related to the incubation period consumed before the cooling rate was changed. Own experiments using cylindrical specimens of 50 mm Dia, made of AISI-4140 steel, have shown that Controllable Delayed Quenching (CDQ) technology has a great potential to increase the depth of hardening, compared to conventional quenching practice. Bending fatigue tests vvith inverse hardened and tempered specimens have shown a significant increase of the fatigue life compared to specimens having normal hardness distribution after quenching. CDQ-technology and 'inverse hardening' can reproducibly be realized using adequate steel hardenability and cross-section size of the workpiece, by quenching in PAG polymer-solution of high concentration. or in high pressure-circulated gases.
Key words: quenching, heat transfer
Raziskane so bile značilnosti porazdelitve trdote na preseku kaljene okrogle palice v odvisnosti od spremembe prenosa toplote na površini palice. Kaljenje z zadržanjem, ki povzroča diskontinuirno spremembo hitrosti ohlajanja, lahko ustvari večjo trdoto v jedru kot na površini. Ta pojav imenovan "inverzna utrditev" sta teoretično razložila Shimizu in Tamura. Odvisen je od kaljivosti jekla, preseka kaljenca in od pogojev kaljenja ter je povezan z inkubacijsko dobo, ki je bila porabljena pred spremembo hitrosti ohlajanja. Naši poiskusi na valjastih vzorcih (J) 50 mm iz jekla AISI-4140 so pokazali, da ima "kontrolirano kaljenje z zadržanjem" velik potencial za povečanje globine kaljenja v primerjavi s klasičnim kaljenjem. Upogibni utrujenostni preizkusi na inverzno kaljenih in popuščenih vzorcih so pokazali pomembno povečanje življenjske dobe v primerjavi z vzorci z normalno porazdelitvijo trdote po kaljenju. CDQ - tehnologija in "kaljenje z zadržanjem" sta lahko reproduktibilna, če ima jeklo primemo kaljivost in je presek kaljenca ustrezen s kaljenjem v polimerni PAG raztopini velike koncentracije ali v visokotlačni krožeči atmosferi.
Ključne besede: kaljenje, prenos toplote
1 INTRODUCTION
The practice of quenching ferrous metals has a very long history, but development of quenching technology vvas first of ali concentrated on choosing the proper quenchant and quenching parameters, i.e. its temperature and possibly the agitation rate. Once this vvas fixed for a specific čase, the heat transfer from the vvorkpiece surface vvas governed solely by the selected quenchant and quenching parameters. Generally the idea prevailed that, for achieving a better through-hardening, a more severe quenching intensity should be applied right at the begin-ning of quenching. If vvorkpieces having bigger cross-section have to be hardened through, the heat extraction from the core is the main problem. Ali efforts vvere concentrated on shortening the quenching time i.e. cooling the core of the vvorkpiece, if possible, belovv Ms temperature before (according to the hardenability of the steel used) the transformation to ferrite and pearlite begins. As a result of applying a severe quenchant, high temperature gradients developed causing high thermal stresses and distortion. Using this approach tvvo factors on vvhich the through hardening depends, have not been taken into consideration:
a)	The cooling rate in every specific point of the cross-section from the austenitizing temperature to the transformation temperature (Ai) is not critical for the microstructure and hardness after quenching. Instead the cooling rate belovv Ai to the Ms point is critical, and different points of the cross-section pass through this temperature range (Ai to Ms) at different times.
b)	When severe quenching intensity is applied right from the beginning of the quenching process, the surface temperature of the vvorkpiece is rapidly decreased to lovv values vvhile the core stili retains high temperature. So, vvhen heat has to be extracted from the core, the temperature difference betvveen surface temperature of the vvorkpiece (T5) and the quenchant temperature (T0) is lovv and according to the Newton's lavv q = a (Ts-T0), the heat extracted is also lovv. This situation can only be im-proved if the heat transfer (characterized by the heat transfer coefficient a (W/m2K)) vvould be increased later during quenching, but this is not the čase in normal quenching practice.
In 1977 the investigation published in USA by Loria1 has shovvn that 'delayed quenching' can in some in-stances increase the depth of hardening, compared to conventional quenching practice. 'Delayed quenching'
means a relatively slow heat transfer from the vvorkpiece surface at the beginning of quenching, followed by a fast cooling with high quenching intensity.
In the same tirne in Japan, Shimizu and Tamura2'3 have given theoretical explanation of this phenomenon stating that it is caused by discontinuous change in cooling rate and the incubation period (at relevant temperature) consumed, before the cooling rate vvas abruptly changed. Latter, experimental investigation by Liščič and Totten4 on one side, and numerical calculation by Chen and Zhou5 on the other side, have shown that at 'delayed quenching' the average cooling rate may be higher below the surface of the vvorkpiece, than at the surface itself. While in čase of normal quenching (vvithout discontinuous change of cooling rate) the cooling rates constantly decrease from the surface tovvards the core, in čase of 'delayed quenching' the cooling rate is lower at the surface (because of mild cooling at the beginning of the quenching process), becoming greater belovv the surface tovvards the core because of the latter abrupt change of heat transfer at the vvorkpiece surface.
Through these vvorks it became evident that the heat extraction dynamics during quenching, and not the quenching time itself is responsible for the hardness distribution on the worpiece's cross-section after quenching.
Studying the pattern of hardness distribution on round bars' cross-section after quenching Shimizu and Tamura3 have introduced the expression of 'inverse' hardening. Opposite to normal hardness distribution it shovvs lovver hardness at the surface and higher hardness in the core.
The experiments have shovvn that the 'inverse' hardness distribution caused by the phenomenon of 'delayed quenching', depends on steel hardenability and on cross-section size of the vvorkpiece. They have also shovvn that (in čase of adequate hardenability and corresponding cross-section size) the 'delayed quenching' has a great potential to increase the depth of hardening, compared to conventional quenching practice.
Chen and Zhou5 state also that 'delayed quenching' can reduce residual stresses and distortion. This state of the art gives, in some instances, the possibility to achieve the biggest possible depth of hardening, simultaneously vvith minimum residual stress and distortion, by a Con-trollable Delayed Quenching.
When quenching in evaporable liquid quenchants, hovvever, the possibility to control a preprogrammed cooling cycle and cause intentionally a 'delayed quench-ing' is very limited, because the only parameter that can be changed during the quenching process is the agitation rate.
Among ali liquid quenchants only the PAG polymer-solutions possess a mechanism that enables them to real-ize a preprogrammed Controllable Delayed Quenching by changing the polymer concentration. The advantage of PAG solution quenchants seams to be that it is possible to achieve the proper balance of the film thickness
and film strength, depending of course on tvvo other parameters, namely the bath temperature and the agitation rate. As it is vvell known higher polymer concentration gives thicker film on the vvorkpiece surface, prolonging the vapor blanket stage i.e. causing a 'delayed quench-ing'.
Recently, Liščič, Grubišič and Totten6 have shovvn that bending fatigue as vvell as impact strength of mechanical components can be increased by 'delayed quenching'.
2 HEAT EXTRACTION DYNAMICS
It is interesting to analyse why 20 years have passed since the phenomenon of 'delayed quenching' vvas first published by Loria and by Shimizu and Tamura until a Controllable Delayed Quenching of real components has been investigated more in details. The ansvver may be found in tvvo follovving reasons:
a)	There vvas no an adequate method to test and re-cord the quenching intensity during quenching in real practice, that could describe the heat extraction dynam-ics. Neither the magnetic quenchometer method, nor the cooling curve analysis of small diameter (12.5 mm Dia x 60 mm length) inconel or silver specimens can describe the heat extraction dynamics vvhen quenching real components.
b)	Only just recent investigations7 have revealed that polymer solutions (PAG) of higher concentration can be used as a quenchant for preprogrammed Controllable Delayed Quenching.
The newly developed Temperature Gradient Quench-ing Analysis System (TGQAS) using the LISCIC/NAN-MAC probe8 of 50 mm Dia x 200 mm length, made of AISI-304 steel, representing a real vvorkpiece, is capable of measuring, recording and evaluation of every quench-ing process in vvorkshop practice, describing the heat ex-traction dynamics by corresponding thermodynamic functions. The probe itself is instrumented vvith three thermocouples at the mid-length cross-section, measuring the temperature at the very surface, 1.5 mm belovv surface and in the center.
Figure 1 shovvs cooling curves recorded in tvvo quenching tests: TEST-1 mineral oil of 20°C vvithout agitation and TEST-26 PAG polymer-solution (UCON-E) of 25% concentration, 40°C bath temperature and 0.8 m/s agitation rate. Figure 2 shovvs calculated heat flux density vs. time betvveen different thermocouple posi-tions.
The characteristic feature in each of quenching tests vvith regard to heat extraction dynamics is the time period from the immersion up to the moment the maximum heat flux density occurs (tqmax).
While for particular oil quenching (TEST-1) tqmax is 14 seconds it is for the described polymer-solution quenching (TEST-26) 72 seconds. The latter one is obvi-ously a 'delayed quenching'.
TESTI.REC	TEST2«.REC
Figure 1: Two quenehing tests recorded by the LISCIC/NANMAC probe (50 mm Dia x 200 mm): TEST-1 - mineral oil, 20°C, without agitation; TEST-26 - PAG po!ymer-soIution (UCON-E), 25%, 40°C, 0.8 m/s agitation rate. Cooling curves for: surface (O), 1.5 mm below surface (□) and center (A)
Slika 1: Dva preizkusa kaljenja registrirana z LISCIC/NANMAC preizkušancem (<|> 50 x 200 mm). Pr. 1 - mineralno olje, 20°C, brez mešanja: Pr. 26 - PAG raztopina (UCON-E), 25%, 40°C, hitrost mešanja 0.8 m/s. Ohlajevalne krivulje za površino (O), 1.5 mm pod površino (□) in sredino (A)
quenching (TEST-26) 72 seconds. The latter one is obvi-ously a 'delayed quenching'.
Because the heat flux density (W/m2) is the real physical measure of the heat extraction, it is interesting to analyse and compare the heat flux density (between 1.5 mm below surface and the surface itself) v.s. time curves for both mentioned tests, shown in Figure 2. For particular oil quenching (TEST-1) only 12.5 seconds, right in the beginning of the quenching process, vvere necessary for increasing the heat flux density from a lovv value of 200 kW/m2 to its maximum of 2600 kW/m2, and 35 seconds vvere necessary for the heat flux density to fall back to 200 kW/m2. For particular polymer-solution quenching (TEST-26), 67 seconds or 5.4 times more vvere necessary for increasing the heat flux density from 200 kW/m2 to its maximum of 2250 kW/m2, but only 23 seconds or 1.5 times less vvere necessary for the heat flux density to fall back to 200 kW/m2.
This analysis clearly shovvs a distinct difference in heat extraction dynamics betvveen the described oil quenching characterized by a fast cooling from the beginning, and the described polymer-solution quenching characterized by a long period of relatively slovv cooling follovved by a sudden change in heat extraction, after burst of the polymer film, vvhich has caused a pro-nounced discontinuous change in the cooling rate, hav-'ng a speciftc influence on transformation behavior of the steel concerned.
The discontinuous change in cooling rate, vvhen quenching in the used polymer solution (TEST-26) can he seen in Figure 1 as a distinct change in the slope of
TESTI-REC	TESTS* .REC
Figure 2: Heat flux densities vs. time for TEST-1 and TEST-26 calculated from recorded cooling curves. Heat flux density betvveen 1.5 mm belovv surface and the surface itself (□); betvveen center and surface (O); betvveen center and 1.5 mm belovv surface (A) Slika 2: Gostota toplotnega toka v odvisnosti od časa za pr. 1 in 26 izračunana iz ohlajevalnih krivulj. Gostota toplotnega toka med globino 1.5 mm in površino (□), med sredino in površino (O) ter med sredino in globino 1.5 mm (A)
the cooling curve for the thermocouple at 1.5 mm belovv surface, at 570°C.
The best way to compare the heat transfer dynamics betvveen both described tests is to compare the amount of heat extracted at different time intervals after immersion of the probe. The curves marked vvith □ in Figure 3 rep-resent the values of the integral (/qdt) belovv the heat
TESTI.REC	TESTS«.REC
Figure 3: Integral (Jqdt) belovv heat flux density curves vs. time for TEST-1, and TEST-26, representing the amount of heat extracted. Heat extracted betvveen 1.5 mm belovv surface and the surface itself (□); betvveen center and surface (O); betvveen center and 1.5 mm belovv surface (A)
Slika 3: Integral (Jqdt) toplotnega toka v odvisnosti od časa za pr. 1 in 26, ki predstavlja ekstrahirano toploto. Toplota ekstrahirana med globino 1.5 mm in površino (□), med sredino in površino (O) ter med sredino in globino 1.5 mm (A)
HEAT EXTRACTION DVNAMICS AT OUENCHING
The Stainiess-steel Specimen of 50 mm Dia x 200 mm quenched tri: Mineral Oil of 20*C - vvithout agitation Cwetting kinetics not included)
88 s	120 s
Color scale
s ni inoanacjasm««
«S50 600 ?5P 7!» 650 «» 850 SOD 450 400 350 3K» 250 200 S5C >C
Time after
Temperature ffefds calculated by a 2-D heat transfer program based on heat transfer coeffiaents obtamed from the USCIC/NANMAC CafcuSatkm program by. B.Smoijan Graphtcat iftferpretatton program by: J.Galinec
immersion:	s
Figure 4: 2-D simulation of temperature fields when quenching the stainless steel specimen of 50 mm Dia x 200 mm in standard mineral oil of 20°C, without agitation. The calculation is based on heat transfer coefficients obtained from the LISCIC/NANMAC probe Slika 4: 2-D simulacija temperaturnega polja med kaljenjem preizkušanca $ 50 \ 200 mm iz nerjavnega jekla v standardnem mineralnem olju in brez mešanja. Izračunano na osnovi koeficientov prenosa toplote pridobljenih z LISCIC/NANMAC sondo
extracted (MJ/m2), vs. time. In oil quenching (TEST-1), the amount of heat extracted starts to rise immediately in third second after immersion, reaching already after 20 seconds a value of 35 MJ/m2. Durtng next 80 seconds it amounted to 55 MJ/m2. In the used polymer-solution quenching (TEST-26) only 5 MJ/m2 vvas extracted until 20 seconds, but in next 80 seconds it reached 87 MJ/m2.
Using the heat transfer coefficient vs. time values calculated from the measured temperatures at the mid-length cross-section of the LISCIC/NANMAC probe, a 2-D heat transfer programme vvas developed for calculat-ing temperature fields during quenching. Figure 4 shovvs a graphical presentation of the cooling a stainless steel specimen of 50 mm Dia x 200 mm length at: 16, 42, 88 and 120 seconds by quenching it in mineral oil of 20°C vvithout agitation (TEST-1). Figure 5 shovvs the same vvhen quenching the specimen in PAG polymer-solution (UCON-E) of 25% concentration, 40°C bath temperature and 0.8 m/s agitation rate (TEST-26). Comparison of Figure 4 and Figure 5 clearly reveals the difference in heat extraction dynamics betvveen those tvvo tests.
It should be emphasized that for transformation kinetics not the cooling rates from austenitizing temperature to Ai, but the cooling rates below Ai, are critical. For the steel grade AISI-4140 e.g. the Ai temperature is 730°C. Analysing the average radial temperature gradi-ents betvveen core and surface in half length cross-section from Figure 4 and Figure 5 respectively, one gets the values given in Table I.
Table I
Austenitizing temperature: 850°C; Radius of specimens: 25 mm				
Time after immersion (seconds)		16	42	88 120
Average temperature gradient betvveen core and	TEST-1	10	12	6 4
surface in half length cross-section °C/mm	TEST-26	2	4	10 6
From the calculated temperature fields in Figure 4 and Figure 5 respectively, and the values given in Table I we can derive the follovving:
In TEST-1 (normal čase of quenching vvith continuous cooling rates), a cooling vvithin the critical tempera-
HEAT EXTRACTION DYNAMICS AT QUENCHING
The Stainless-steel Specimen of 50 mm Dia x 200 mm quenched in: PAG Polymer-solution of 25% concentration; 40°C temp.; 0.8 m/s agitation rate (vvetting kinetics not included)
Figure 5: 2-D simulation of temperature fields by quenehing the stainless steel specimen of 50 mm Dia x 200 mm in PAG polymer-soIution (UCON-E) of 25% concentration; 40°C; 0.8 m/s agitation rate. The calculation is based on heat transfer coefficients obtained from the LISCIC/NANMAC probe
Slika 5: 2-D simulacija temperaturnega polja pri kaljenju preizkušanca <f> 50 x 200 mm iz nerjavnega jekla v PAG polimerni raztopini (UCON-E) s koncentracijo 25%. 40°C in hitrostjo mešanja 0.8 m/s. Izračunano z uporabo koeficientov prenosa toplote pridobljenih z LISCIC/NANMAC sondo
88 s	120 s
Color scaie
probe
■■^□□□□[□□□IM«™
850 800 750 700 650 600 550 500 450 400 350 300 250 200 150°C
Time after
immersion: 16s	42 s
Temperature fields calculated by a 2-D heat transfer program based on heat transfer coefficients obtained from the LISCIC/NANMAC Calculation program by: B.Smoljan Graphical interpretation program by: J.Galinec
ture range (700°C to 400°C) for the core, between 42 and 88 seconds is obtained with a decreasing temperature gradient, i.e. decreasing heat extraction flux from the core to the surface. Once the surface temperature has fallen to lovv values (about 200°C after 88 sec.), the heat transfer has decreased very much, because of the small temperature difference betvveen the vvorkpiece surface and the surrounding fluid. This heat extraction dynamics results in the normal hardness distribution i.e. substan-tially lovver core than surface hardness.
In TEST-26 (a delayed quenching vvith discontinuous change of cooling rates), cooling of the core from 42 to 88 sec. (i.e. betvveen 750°C and 600°C) is obtained vvith increasing temperature gradient, i.e. tncreasing the heat extraction flux from the core to the surface, resulting in inereased core hardness.
3 TRANSFORMATION KINETICS WHEN DISCONTINUOUS CHANGE OF COOLING RATE OCCURS
From the moment the austenitized vvorkpiece is im-mersed in the quenching fluid, two different processes start: the thermodynamic process of heat extraction and the metallurgical process of structure transformation. The latter one starts actually in different times for different points of the eross-seetion, vvhen the temperature in each point falls to A|. These times depend on the cross-seetion size and the cooling intensity of the quenching fluid. The resulting hardness in a particular point de-pends on constituents of the structure transformed, vvhich depend heavily on the hardenability of the steel concerned i.e. on ineubation times at every isotherm. Because ineubation times are counted only at temperatures belovv Ai, for each particular point of the eross-seetion the cooling rate in the critical temperature range A/ to Ms, is of paramount importance.
Shimizu and Tamura2 have found that the pearlitic transformation behavior vvith cooling rates diseontinu-
ously changed during cooling was different from that given by an usual CCT diagram, and that this transfor-mation is related to the incubation period consumed be-fore changing the cooling rate. In čase of delayed quenching some of the incubation period is consumed at the surface of the vvorkpiece, while it is not at the center. The incubation period at any given isotherm is the time until the transformation starts (Z), while (X) is the incubation period consumed before the discontinuous change of the cooling rate has taken plače. Figure 6 which is a schematic illustration of delayed quenching, shows that at time tj and temperature Ti (point P) a discontinuous change of cooling rate occurred. Up to this moment the surface of the vvorkpiece has consumed a share (X) of the total incubation time (Z), but the center has not, be-cause at the moment ti the center had a temperature above Ai. Further cooling belovv the point P has pro-ceeded vvith substantially increased cooling rate, changing the transformation start curve as shovvn in Figure 6. Because for the center no incubation time has been consumed, the cooling curve for center starts from temperature Ai at zero time! In this way the cooling curve for center, vvhich doesn't intersect any pearlitic region, results in higher hardness than the cooling curve for the surface vvhich has started from the point P and inter-sected a portion of pearlitic region. This is the theoretical explanation of 'inverse' hardness distribution.
b)
A,*—-
\	/^Pearlitic transformation starts
£ ft Yj (originated at Ai)
~ Yv V7— Pearlitic transformation starts 2 \\ N (originated at P)
E X\l /V"Surface " f \rrA
Center
Time
Figure 6: Schematic illustration how delayed quenching causes inverse hardening, according to2
Slika 6: Shematična predstavitev kako kaljenje z zadržanjem ustvari inverzno utrditev. Po viru 2
4	HARDNESS DISTRIBUTION AFTER QUENCHING AND AFTER TEMPERING
Figure 7 shovvs the normal hardness distribution measured across the section of a 50 mm dia bar of AISI-4140 after quenching in ordinary mineral oil of 20°C bath temperature vvithout agitation (curve No 1), and the inverse hardness distribution measured after quenching the same bar in the PAG polymer solution (UCON-E) of 25% concentration, 40°C bath temperature, and 0.8 m/s agitation rate (curve No 2). This shovvs the high capacity of delayed quenching technique to influence the depth of hardening.
Because low-alloyed structural steels (like the AISI-4140) are used in hardened and tempered condition, it is of interest to see hovv a normal and an inverse hardness distribution curve, respectively, look like after temper-ing. Figure 8 shovvs this for specimens used to plot the hardness distribution curves in Figure 7, after tempering at 480°C for 2 hours. Normal hardness distribution (curve No 1) has retained the same shape as after quenching, vvhile the inverse hardness distribution (curve No 2) gave a uniformly distributed hardness over the cross section. This is result of the knovvn fact that at tempering the higher hardness values decrease more than the lower hardness values. The hardness difference at the center of about 6 HRC indicates that inverse hardness distribution guaranties after tempering a structure of tempered martensite in the core, vvhile in čase of normal hardness distribution, besides tempered martensite other (softer) structure constituents are present in the core. With regard to mechanical properties, as it is vvell knovvn (especially for high strength levels), that tempered fine-grained martensite yields the highest toughness of ali mi-crostructures.
5	INFLUENCE OF HARDNESS DISTRIBUTION ON FATIGUE PROPERTIES
For bending fatigue tests9 especially designed specimens of 300 mm length vvith the critical diameter of 50 mm vvere machined of the same heat of the American made steel AISI-4140. Ali specimens vvere austenitized in a protective atmosphere to 860°C for 80 minutes. The specimens having normal hardness distribution vvere quenched one by one, vertically, in used mineral oil of 20°C vvithout agitation. The specimens having inverse hardness distribution vvere quenched in PAG polymer-so-lution (UCON-E) of 25% concentration, 40°C bath temperature and 0.8 m/s agitation rate. After quenching ali specimens have been tempered in a vacuum furnace at 500°C for 2 hours.
Figure 9 shovvs the used test rig for bending fatigue tests. Ali tests vvere performed in normal environmental conditions under a specific load programme. One round of the used load programme consisted of 7,000 cycles vvhich vvere subdivided into tvvo parts: - 5,000 load cycles vvith the regular load amplitude
AISI-4140 Batch No 7345$
3/4/? 1/2 R 1/4/? 0 1 UR MIR 3/4/? _ _50 mm Dia.
Figure 9: Test rig for bending fatigue tests Slika 9: Priprava za preizkus upogibne utrujenosti
Figure 7: Hardness distribution curves measured on the cross-section of a 50 mm Dia x 200 mm cylinder inade of AISI-4140, after quenching under the following conditions:
1)	Mineral oil of 20°C, without agitation
2)	PAG polymer-solution (UCON-E) of 25% concentration; 40°C bath temperature and 0.8 m/s agitation rate
Slika 7: Porazdelitev trdote po preseku preizkušanca (J) 50 x 200 mm iz jekla AISI-4140 po kaljenju v naslednjih pogojih:
1)	mineralno olje, 20°C, brez mešanja
2)	PAG polimerna raztopina (UCON-E), 25%, 40°C in hitrost mešanja 0.8 m/s
- 2,000 load marker cycles vvith 25% higher than the regular load amplitude.
The 2,000 load marker cycles vvere used to obtain in-formation about the crack grovvth rate and a possible influence of the hardness distribution on it. The informa-tion about the crack grovvth rate is expressed in form of share of the crack grovvth phase in the total test Iife:
o a: x
vi rt 41 C T) i— O
r
Figure 8: Hardness distribution after tempering at 480°C for 2 hours:
1)	Specimen of 50 mm Dia x 200 mm, quenched in mineral oil of 20°C, vvithout agitation
2)	Specimen of 50 mm Dia x 200 mm, quenched in UCON-E: 25%; 40°C; 0.8 m/s
Slika 8: Porazdelitev trdote po popuščanju 2 uri pri 480°C:
1)	Preizkušanec <j> 50 x 200 mm kaljen v mineralnem olju pri 20°C brez mešanja
2)	Preizkušanec <|> 50 x 200 mm kaljen v UCON-E 25%, 40°C, 0.8 m/s
NrNc Nf
in %, vvhere Nf is the number of cycles of the total test life and Nc is the number of cycles to the initial crack. During the tests the cylinder displacement and the load amplitude vvere recorded to determine Nc and Nf values. The Nc value refers to the beginning of the stiffness loss of the specimen due to initiated crack.
The fatigue tests vvere performed on different loading levels under pulsating sinusoidal Ioads vvith the fre-quency of f = 16 Hz and a stress ratio R = Fmin/Fmax = 0 vvhich led to the nominal stress amplitudes in the critical area of the specimens. The test results represented by fatigue life to the initial crack versus the nominal stress amplitude (S - N curves) are shovvn in Figure 10.
Summarising these results it can be concluded, (al-though the number of tested specimens vvas lovv for a statistically confirmed data), that an increased fatigue life vvas achieved vvith specimens having inverse hardness distribution compared to specimens having normal
Moltria! IZCrMflt (AISI-lllO) K, • t, 85
Stltlf - Rnlia R.^^.O
100-1-1--1 I I I I I- ---1111 ' ---- 1 1 1 ' '
Iti	t«S
Number §1 crclM |N)
Figure 10: Bending fatigue test results of specimens vvith normal and vvith 'inverse' hardness distribution after quenching (both tempered to
500°C for 2 hours)
Slika 10: Rezultati preizkusov upogibne utrujenosti preizkušancev z normalno in z inverzno porazdelitvijo trdote po kaljenju (oba sta bila popuščena 2 uri pri 500°C)
35-1_i_l_i_l i_i_i_i_i_i_i
0 2 4 6 8 10 12 14 16 18 20 22 24 26 Distance (rom the surface (50 mm Dia) /mm
AISI-4140 Batch No 73456
hardness distribution. At the stress level of 270 MPa, at which most tests have been performed, this increase is expressed by a factor of about 7. The crack propagation phase, compared to the total fatigue life vvas more uniform for specimens vvith inverse hardness distribution and amounted to 13 to 20%, depending on the stress level. Additional fatigue tests are planned to increase the statistical validity of the achieved results.
6 CONCLUSION
The above described investigation shovvs that the hardness distribution on the cross-section of the vvorkpiece after quenching can be influenced and greater depth of hardening and better mechanical properties can be achieved by a predetermined and controllable heat transfer dynamics. In future, therefore, the quenching technology vvill most probably adopt the control of heat transfer from the surface of the vvorkpiece instead of let-ting it occur by itself (depending only on the quenchant and quenching parameters selected), as in today's practice. If so, the question vvill arise: By vvhich means a controlled heat transfer at quenching is possible?
For liquid, evaporable quenchants (as the hitherto in-vestigations shovv), this is possible by using polyal-kylene-glicol (PAG) polymer-solutions of sufficiently high concentration of adequate temperature and agitation rate. In gas quenching applications (especially in vacuum furnaces vvith pressurized high velocity gases), more time is available during quenching than in čase of liquid quenchants, to change the main cooling parameters i.e. the gas pressure and gas velocity.
In order to find out vvhether in a particular čase the workpiece's cross-section size and hardenability of the steel grade in question are suitable for quenching with controlled heat extraction dynamics and to optimize the relevant quenching parameters the computer simulation vvill be necessary.
The base for such a simulation are the tvvo follovving requirements:
- The CCT-diagram of the steel-grade in question, vvhich characterizes its hardenability and allovvs to
overiay the calculated cooling curves for different points on the cross-section, to evaluate the transformation kinetics. - Heat transfer coefficient values a = f(t) (W/m"K) betvveen the vvorkpiece surface and the quenching medium for the vvhole quenching process, character-izing the changes in quenching intensity, vvhich allovvs to calculate the relevant cooling curves in every cross-section point of different bar diameters. To get relevant heat transfer data, a vvorkshop de-signed method to measure and record the quenching in-tensity of different quenchants, as described in8, is re-quired.
7 REFERENCES
1	E. A. Loria: Transformation Behavior on Air Cooling Steel in A3-A1 Temperature Range, Metals Technology, (1977) october, 490-492
2	N. Shimizu and I. Tamura: Effect of Discontinuous Change in Cooling Rate During Continuous Cooling on Pearlite Transformation Behavior of Steel, Transactions /SIJ, 17 (1977) 469-476
3	N. Shimizu and I. Tamura: An Examination of the Relation Between Quench-hardening Behavior of Steel and Cooling Curve in Oil, Transactions ISIJ, 18 (1978) 445-450
"B. Liščic, G. E. Totten: Controllable Delayed Quenching, Proceedings of the International Heat Treating Conference Equipment and Processes, April 1994, Schaumburg, Illinois, USA, 253-262 5 M. Chen and H. Zhou: Numerical Heat Transfer Analysis on the Effect of Enhancing the Thickness of the Hardened Layer by Delayed Quenching, Jinshu Rechuli Xuebao (Transactions of Metal Heat Treatment), 14 (1993) 4, 1-6 (in Chinese) 6B. Liščic, V. Grubišic' and G. E. Totten: Inverse Hardness Distribution and its Influence on Mechanical Properties, Proceedings of the 2nd International Conference on Quenching and the Control of Distortion, 4-7 November 1996. Cleveland, Ohio, 47-54
1 B. Liščic: Investigation of the Correlation Betvveen Polymer-Solution (PAG) Concentration and Inverse Hardening Distribution Curves, Internat Report No 11/92 of Laboratory for Heat Treatment, Faculty of Mech. Engineering and Naval Architecture, Zagreb, March 1992 8B. Liščic, S. Švaič and T. Filetin: VVorkshop Designed System for Quenching Intensity Evaluation and Calculation of Heat Transfer Data, Proceedings of the Ist International Conference on Quenching and Control of Distortion, 22-25 Sept. 1992, Lincolnshire, Illinois, 17-26
9 Test Report Nr 7710, 24 Nov 1994 from the Fraunhofer Institut fur Betriebsfestigkeit, Darmstadt, Germany