HAL Id: hal-02056160
https://hal-mines-albi.archives-ouvertes.fr/hal-02056160
Submitted on 14 Mar 2019HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Heating a first approach
Yannick Le Maoult, Fabrice Schmidt, V Laborde, Mouna El-Hafi, Philippe
Lebaudy
To cite this version:
Yannick Le Maoult, Fabrice Schmidt, V Laborde, Mouna El-Hafi, Philippe Lebaudy. Measurement and Calculation of Preform Infrared Heating a first approach. 4th International Workshop on Advanced Infrared Technology and Applications, Sep 1997, Florence, Italy. �hal-02056160�
MEASUREMEN TAN DCALCULATION OFPREFORMIN FRARED HEATIN G:AFIRSTAPPROACH
Y. LE MAOULT (*), F. M. SCH MIDT (*),
V. LABORDE (*) , M. EL H AFI (*) , P. LEBAUDY (**),
Abstract
In a first step , the goal of the w ork is to obtain an accu rate characterisation of the heat sou rce of the infrared em itter. In a second step , the com p u tation of the view factor betw een the lam p and the p reform has been p erform ed u sing a Monte-Carlo m ethod . Then, an exp erim ental valid ation of the view factor com p u tation u sing qu antitative infrared therm ograp hy has been d evelop ed . This w ork is a first contribu tion to the stu d y of rad iative heat transfer w hich are involved in infrared heating of p reform related to the injection stretch -blow m old ing of therm op lastic bottles.
N omenclature
Bi Biot nu m ber total Absorp tivity of a sheet
cp Sp ecific H eat of cop p er t Em issivity of tu ngsten
F12 View factor Wavelength
I Intensity of the cu rrent H eat flu x
m m ass of a sheet Resistivity of tu ngsten
Nt Total generation of c Density of cop p er rand om nu m bers
R Resistance Variance
S Su rface of a sheet Therm al tim e constant
T Tem p eratu re k therm al cond u ctivity of cop p er U Su p p ly of a lam p h global heat transfer coefficient V Volu m of a sheet
(*) ÉCOLE DES MINES d'ALBI, Campus universitaire Jarlard, Route de Teillet, 81013 ALBI CT
CEDEX 09 (FRANCE)
1. Introduction
Form ing p rocesses of therm op lastic su ch as PET, PVC or PP (injection stretch -blow m old ing, therm oform ing,..) need a heating stage [1, 2]. A tu be -shap ed or sheet-shap ed p reform is heated in an infrared oven above the glass transition tem p eratu re (for PET Tg # 80 °C) in ord er to p rovid e a ru bber -like state before the inflation step . The op tim isation of the heating stage is cru cial. The final thickness d istribu tion of the p rod u ct is d rastically controlled by the initial tem p eratu re d istribu tion insid e the p erform . Besid es, the op tim isation of the oven w ill p erm it to m inim ise the cost of energy. So the p relim inary w ork p resented here d eals w ith tw o p articu lar p oints :
- The characterization of the infrared lam p u sed to heat the p reform (tem p eratu re, sp ectral, and d irectionnal asp ects)
- The interaction betw een the lam p and the p reform (view factor), this last p oint is n ot obviou s w hen the em ission of the lam p is strongly d irectionnal and w hen the rad iant energy interact w ith a sem i- transp arent m ed ia [3].
A typ ical oven w ith blow ing p rocess is show n in figu re 1
2. Characterization of the heat source : the infrared emitter
2.1 p hysical d escrip tion of the lam p
w e u sed a p hilip s lam p w hich is com m only p lu gged in ind u strial ovens. The nom inal p ow er of this sou rce is 300 W. The d ifferent com p onents of this lam p are :
- a sp ecificic d ou ble sp iraled filam ent contained in a tu bu lar enclosu re in qu artz.
- A reflector (p olished alu m iniu m ) fitted on the back sid e of the qu artz, its m ain role is to increase the efficiency of the em itter in the forw ard d irection.
A view of the lam p is show n in figu re 2 :
figure 2
2.2 Tem p eratu re of the tu nsten filam ent.
An electrical setu p has been realized to m ake tem p eratu re m easu rem ent of the filam ent. A variation of the su p p ly U of the lam p is necessary to p lot a cu rve of U versu s I(am p ), this grap h is easily converted to a list of valu e of the resistivity (T) from a w ell-know n relationship
R(T) (T).L(T)
Where L and S are the length and the cross section resp ectively. The variation of the term L/ S versu s T is very w eak com p ared to (T) and can be assu m ed as a constant, then an inverse p olynom ial interp olation based on the d ataset of r from [4] give T = f() w ith T in Kelvins :
T = -5.723e-22
+ 36.58+115.3 (2)
at nom inal Pow er (P=300 W), w e find : T = 2030±50°C
This ap p roach has been also valid ated w ith an op tical p yrom eter. The absolu te u ncertainty of 50 °C is not a d eterm inant factor becau se of a w eak
variation of em issivity betw een 2000 and 2100 °C.
2.3 Sp ectral em issivity
In ord er to m od elize the sp ectral intensity em itted by this kind of lam p , w e have to know accu rately the sp ectral em issivity of the tu ngsten w ire and the transm ittivity of the qu artz tu be. This can be d one by u sing sp ectral d at a of tu ngsten [5].Unfortu nately, to fit a com p lete sp ectra at the right tem p eratu re, (beyond 2.5 µm ), it w as necessary to u se the Dru d e ap p roach w hich is based on electrom agnetic theory. The p articu lar exp ression of this m od el is :
(,T )= 36.5((T ) )0.5464 (T )
(3)
This relation becom es accep table w hen the w avelength > 3 µm . The figu re 3 show s the the com p ilation of d ifferent characteristics, esp ecially intensity inclu d ing the effect of the qu artz w ind ow :
2.3 Directional em ission
A very im p ortant p oint in oven d esign : an IR sou rce w ith a very strong
d irectionality w ill p rod u ce overheating (hotp oints) on the p reform (w e have to keep in m ind that p olym ers have a very low therm al cond u ctivity, a typ ical valu e is # 0.2 W/ m .K). Another sp ecific exp erim ental setu p has been u sed to stu d y this p roblem : a Si(IR) sensor (p hotod iod e) d escribes a half circle arou nd the lam p and give the integrated intensity for several d irections. The sp ectral resp onsivity of su ch op tical com p onent is lim ited to the sp ectral d om ain 750 to 1100 nm . The resu lts are p resented on figu re 4 for d ifferent cases. It ap p ears that the rep rod u ctibility of the sp atial d istribu tion of intensity is affected after reassem bling the reflector.
This p oint highlights the fact that an op tical op tim ization is necessary to heat p rop erly a p rod u ct:
0 0,2 0,4 0,6 0,8 1 1,2 1,4 -0,5 0 0,5 qualitativ e emiss ion with reflector without reflector with reflector (after reassembling) Lambertian (ideal case) 90° 0° 45° 45° 90° figure 4
3. Interaction betw een the lamp and the preform : Computation of the view factor (first approach : lambertian case)
3.1 Assu m p tions : the su rfaces are gray and u niform . These su rfaces are consid ered as d iffu se em itter s and the geom etrical configu rations are sim p le (rectangu lar sheets) as show n below (figu re 5) :
O X Z Y S1 S2 a(1) b(1) a(2) b(2) zo xo Emitter : size of the lamp
Receiver the “preform” P2 P1 n(P1) 1 2 r n(P2) n figure 5
The know led ge of the view factor F12 (d im ensionless p aram eter) is a fu nd am ental p oint in rad iative transfer to com p u te the therm al p ow er absorbed by the receiver (2) from the em itter (1); so w e have :
abs F12 emis (4)
bibiliograp hic search [6 / 7] show s that, excep t for sim p le cases w here analytical solu tions are available, m ost configu rations m u st be treated w ith nu m erical m ethod s. As w e show , the Monte Carlo m ethod has been chosen for this test compared to Nusselt’s sphere results. The principle of Monte Carlo m ethod is based on statistical ap p roach ap p lied to integral calcu lation. A generation of rand om nu m bers is p erform ed accord ing to a p articu lar
d istribu tion fu nction. The im p lem entation ru n s on a variou s set of d ata and the p rogram generates p oints belonging to the d ifferent su rfaces (the d istibu tion can be isotrop ic or not for each su rfaces). At least the estim ation of the error is m ad e by variance calcu lation. The resu lts are gathered in a table w hich p resents a com p arison w ith analytical solu tion if p ossible or another nu m erical m ethod s. We tested the follow ing sizes for the com p u tation : Em itter : (8 cm * 1 cm ) / Receiver (3 cm * 20 cm ). At first, w e have tested a non -centered configu ration (Receiver shifted resp ect to the Em itter). The com p arison w ith an analytical solu tions show s that the relative error rem ains w eaker than 2 %; bu t a m ore interesting case is the follow ing : configu ration w here the receiver is inclined resp ect to the em itter (no analytical solu tion available):
(°) x0 z0 d F12 N u sselt’s sp here F12 Monte Carlo rel. error in % 15 0 0 10 0,12034 0,12169 1,1 15 0 0 20 0,04017 0,04005 0,3 30 0 0 10 0,12677 0,12709 0,2 30 0 0 20 0,03918 0,03818 2,6 15 5 5 10 0,07455 0,06482 13 15 8 0 10 0,05043 0,04950 1,8 15 0 5 10 0,11802 0,09765 17,3
The nu m ber of elem ents for the em itter (N u sselt m ethod ) is 150 *150 and N t (statistical generations for Monte Carlo) is 3000 . The Agreem ent is good excep t in tw o cases w here a hid d en su rface p roblem ap p ears.
4. Experimental validation of the view factor computation using quantitative infrared thermography.
4.1 Descrip tion of the exp erim ental setu p (figu re 6) :
The p reform is rep resented here by a sheet of cop p er coated w ith a flat black p ainting (Em issivity = 0.92 betw een 0.5 and 20 m ) and the heater is assem bled w ithou t reflector at nom inal p ow er. The cam era u sed in this exp erim ent is a 880 LW (8-12 m band ) / AGEMA. The frequ ency of analysis is equ al to 25 fram es/ s and the d evice is p lu gged on a 12 bits d ata acqu isition board d rive by a real tim e softw are (PC AT).
4.2 Sim p lified m od el of heat transfer in the cop p er sheet
Exp erim ental p roced u re : the cam era observes the back face of the cop p er sheet w hich has been screened from the rad iant sou rce w ith a p olished alu m iniu m p late. The screen is su d d enly rem oved (H eavisid e step of flu x on the sheet) and the tem p eratu re evolu tion of the sheet is p lotted u p to the stead y state. We m ad e the assu m p tion (w hich is valid ated at the end ) that the heat transfer in the sheet can be treated as a lu m p cap acitance m od el (transient asp ect) so Bi m u st be < 0.1. The averaging equ ation is :
c Vcp dT
dt 2 hgS T
Ta
abs (5) and øabs is d ed u ced from (4)The p ow er em itted by the lam p is know n by the electrical m esu rem ents of U, I so for the stead y state Pelec = em is. = U.I, the solu tion of (5) is
T Ta abs 2 h g S 1 e t
(6)
the tim e constant of a sheet is equ al to mcp / 2hg.S and Ta the initialtem p eratu re is equ al to the am biant tem p eratu re. The tem p eratu re evolu tion of the sheet is p lotted hereafter (figu re 7) :
time
FIGURE 7 :
temperature versus time of the sheet
T
Tf
Ta
initial slope =
abs/mcpF12 = 2.
abs UI
Three configu rations w ere tested (resp ect to the heater) : centered sheet, shifted sheet, inclined sheet. A therm ogram is show n on figu re 8 :
selection of the zone figure 8
The zone N °1 has been chosen as a reference for the com p u tation of the view factor (this zone is isotherm al at 3 %). The resu lts are listed below :
Results
distance lamp /sheet =10 cm
centered sheet shifted sheet = 5 cm inclined sheet 15 ° Ta(°C) 28,9 28,2 29,9 Tst(°C) 76,3 68,5 79,2 (seconds) 210 219 217 hg W/ m 2 K
16,9 16,1 16,3 abs
W 9,32 7,14 9,81 F12 0,083 0,063 0,087the valu e of the Biot nu m ber is # 8.10-5 and the relative error rem ains very low (<5 %). The agreem ent betw een the sim u lations and the exp erim ental resu lts is fair :
centered sheet shifted sheet inclined sheet F12 (experimental) 0,083 0,063 0,087 F12 (numerical) 0,0866 0,062 0,0913 relative error in % 4,2 1,7 4,7
Conclusions and prospects
A first ap p roach of the characterization of the interaction betw een an infrared heating system and a p reform has been p resented . The sp ectral and geom etrical asp ects have been d iscu ssed for a fu rther ap p roach; and a sim p le therm al m od el to p erform a m easu rem ent of the view factor in several configu rations is p resented and easily valid ated w ith the infrared therm ograp hy techniqu e (lam bertian case). The m ost p articu lar p oint of this p ap er is the u sing of the Monte Carlo m ethod ; the m ain interest of this « p hysical com p u tation » (in ou r context) can be su m m arized as follow s :
- the m ethod w ork w ith com p lex geom etry
- a p recise d escrip tion of the rad iative transfer in a sem i-transp arent m ed ia su ch as Polym ers (PET/ PVC..) can be d one
- The introd u ction in the m od el of an anisotrop ic d istribu tion of intensisty is p ossible (a sp ecific lam p w ith reflector for exam p le).
N ow and for next m onths : The com p u tation of the view factor in the anisotrop ic case is being tested and a m ore accu rate exp erim ental setu p has been d evelop ed to stu d y the heat flu xes on a p lastic p late (this p late is instru m ented w ith therm ocou p les). This d evice w ill be u sed to d e scribe m ore realistic situ ations closer to the ind u strial context.
References
[1] G.D enis, Extru sion-sou fflage et injection -sou fflage avec ou sans bi-étirage,
Techniques de l’ingénieur, A3700,1989.
[3] P. Lebaudy, Etu d e et sim u lation d e la rép artition d es tem p ératu res d ans
u n cylind re creu x d e PET sou m is à u n rayonnem ent infrarou ge, thèse d e d octorat, 1989.
[4] Smithells Metals, Reference hand book - Brand es book, 1990.
[5] F. D esvignes , Rayonnem ents op tiqu es - Masson - 1991.
[6] Siegel R. and J.R How ell, Therm al Rad iation H eat Transfer, Mc Graw H ill,
N ew York, 1981
[7] G. Ritoux , Evalu ation nu m ériqu e d es facteu rs d e form es, Revu e d e