Heat transfer in an autoclave for processing thermoplastic composites P.F. Monaghan, M.T. Brogan and P.H. Oosthuizen

This paper deals with a preliminary analysis of the heat transfer within autoclaves as used in the diaphragm forming technique. Simple experiments are carried out to estimate the convection and radiation heat transfer and a simple mathematical model is developed. The results presented for the autoclave tested show that the convection coefficient does not vary significantly along the length of the autoclave and radiation may be as high as 60% of the total heat transfer. The temperature distribution through a mould and lay-up is measured and this shows that the thermal resistances inside and outside the mould and lay-up are both significant during the process.

Keywords: autoclaves; heat transfer; thermoplastic composites; diaphragm forming; model; experiment

NOMENCLATURE

A area, m2 CP specific heat of plate, J kg-’ K- ’ F view factor h convection coefficient, W m-’ K-’

J number of time steps radiosity of surface, W mm2

k thermal conductivity, W m- 1 K- ’ L characteristic length, m m mass of plate, kg N number of surfaces within enclosure 4 cOnv convection heat transfer rate, W &ad radiation heat transfer rate, W 4 tot total heat transfer rate, W Tair autoclave air temperature, “C

Tp average experimental plate temperature, “C

it distance from door of autoclave, m time step

E surface emissivity

Subscripts

i surface i j surface j conv convection P plate rad radiation

side wall sl; duct plate dr door he heat exchanger

INTRODUCTION

General

During the last decade the arrival of a new generation of high performance thermoplastic composites has stimulated the development of manufacturing processes for these new materials. The main manufacturing techniques for thermoplastic composite materials are bag moulding, pultrusion, thermoforming and injection moulding. Research at the Department of Mechanical Engineering at UCG (University College Galway) 1,2 has concentrated on diaphragm forming techniques within an autoclave. These techniques are relatively new and require substantial development and research to make them both technically and commercially viable.

The trial-and-error approach adopted in most composite processing techniques gives little understand- ing of the various parameters involved in the process and contributes to the current high cost of composites3. One of the obstacles to the commercial viability of diaphragm forming techniques is the length of the process cycle times. The majority of this process cycle time is used in the heating and cooling of the mould and lay-up. As a first step in reducing the heating and cooling time, the flow field within autoclaves and the heat transfer character- istics of autoclaves, moulds and lay-ups must be fully understood. A full understanding of the heat transfer characteristics of autoclaves and diaphragm forming techniques will also lead to the prediction of the physical and mechanical properties of thermoplastic composite components.

0956-7143/91/03-04233-10 0 1991 Butterworth-Heinemann Ltd

Composites Manufacturing Vol 2 No 3/4 1991 233

Review Dwell Form Consolidate Coolina

Ogale and McCullough4*’ show that the physical and mechanical properties of the thermoplastic composite are influenced by the cooling rates of the parts during processing. O’Bradaigh and Mallon have shown that processing composite materials at temperatures outside the optimum process temperature range also affects the mechanical and physical properties of the end product.

In spite of its importance, very little research has been carried out on the heat transfer in autoclaves. Ghariban, Haji-Sheikh and Lou7 have measured the convection heat transfer coefficients over a flat plate, the air turbulence levels and the local velocities, using three channel hot-wire anemometry, within a scaled down physical model of a concentric cylinder autoclave. They show that the ratio of the experimentally measured convection coefficients to that theoretically predicted, varies from near 3 at the entrance region to 1 downstream within the model autoclave and that the turbulence levels are very high along the entire length of the autoclave. Murphy* indirectly estimated an overall surface heat transfer coefficient over a mould and lay-up by modelling the heat transfer through the mould and lay-up as a 1-D problem, using finite element analysis (FEA). Murphy* and Ghariban et ~1.~ do not attempt to determine the relative importance of the different modes of heat transfer within autoclaves.

The long-term objectives of a project undertaken by the authors of this paper are to develop an understanding of the heat transfer mechanisms involved in diaphragm forming within autoclaves and then develop a general model which will assist design and process engineers in

a

~- ring ylClamping

e,MNMMHdTop diaphragm

/ Prepreg

Vacuum ring

4M1~~I~~,~tom diaphragm

Mould

Vacuum

I- I I I I I I 0 10 20 30 40 50 60

Process time (min]

Figure 2 Schematic diagram of a typical process cycle for APC-2 showing the temperature and pressure curves against time

making decisions which maximize heat transfer and minimize the process cycle times. In his general review of the subject Guceri3 summarizes the role of the thermal engineer by stating, ‘In thermoplastic-matrix composites the single most important factor, therefore, becomes the rate of application and subsequent removal of heat from the composite’.

Diaphragm forming

In diaphragm forming in an autoclave, the unconsoli- dated composite plies and a vacuum ring are held between two plastically deformable sheets known as disphragms (see Figure 1 (a)), which may be metallic or polymeric. The diaphragms are clamped to the mould and the air is evacuated from between the diaphragms. The entire assembly is heated in the autoclave and formed, under pressure, into the mould shape (see Figure 1 (b))). Figure 2 shows a schematic of a typical process cycle for APC-2 (continuous fibre-reinforced PEEK). At present, complete process cycle times are in the range 30-60 minutes. Unlike thermosets, thermoplastics do not have to be cured for long periods at different temperatures and can be heated rapidly to the processing temperature.

The specific aims of this paper are (a) to estimate the energy flows to moulds within the UCG experimental autoclave and (b) to estimate the thermal resistances to heat transfer through a mould and lay-up heated within the experimental autoclave. The relative importance of the different modes of heat transfer may then be established.

P+ HEAT TRANSFER MECHANISMS

This section describes the different modes of heat transfer that occur in this process.

Heat transfer within the autoclave

The two different modes of heat transfer to moulds within autoclaves are :

b 1) Forced convection heat transfer to the mould from

Figure 1 (a) Exploded view of a typical mould and lay-up assembly used in the diaphragm forming process; (b) schematic of forming process showing the lay-up being forced into the cavity of the mould (both courtesy of M. Monaghan’ )

the air circulating through the autoclave working section.

2) Radiation heat transfer between the interior surfaces of the autoclave working section and the mould.

234 Composites Manufacturing No. 314 1991

The relative importance of convective and radiative heating depends on the autoclave design and the processing cycle used. Autoclaves are generally cylin- drical pressure vessels which are divided into a central working section, where the mould is placed, and a ducting system for air circulation. The air is drawn by a fan through the working section and the ducting system. In commercial autoclaves, a number of different ducting designs exist, e.g. the annulus section formed between the pressure vessel wall and a concentric cylinder placed inside the vessel (Scholz’ ), or the segment of the circular cross-section formed by a flat metal plate resting in the bottom of the autoclave ( LBBC9). Autoclaves are normally heated by electric resistance heaters. The relative importance of radiative heating of the mould depends on the location of these resistance heaters. In typical autoclaves, the authors have identified three such locations :

1) Band heaters in the working section walls’. In this case both the mould and working section walls receive radiant heat direct from the resistance heaters.

2) Resistance heaters placed in the air ducts which are located parallel to the working section’. In this case some of the working section walls receive radiant heat direct from the resistance heaters.

3) Resistance heaters placed outside the working section and ducts’. In this case, there is no (or very little) direct radiant heating of either the mould or working section walls by the resistance heaters.

In this paper the UCG autoclave’ is studied. Compared to other autoclaves**9 the radiative contribution to heat transfer of the mould in the UCG autoclave is expected to be relatively high. Both preheat and non-preheat processing cycles are considered.

The convective heat transfer within autoclaves depends mainly on the different autoclave geometries, mould shapes and sizes and the air velocity. The radiative heat transfer to moulds depends mainly on the view factors between the mould and autoclave surfaces and the

Heat transfer to top surface of lay-up

Diaphragm

Heat transfer due to conduction and across contact resistances

1

Composite Prep9

Diaphragm

II radiation, conduction and natural convection II

Air gap

Heat transfer due to conduction

. I

Mould

Heat transfer to bottom surface of mould

Figure 3 Schematic diagram of heat transfer mechanisms through a typical mould and lay-up

emissivities of the mould and autoclave materials. The autoclave and mould surfaces are usually made from either mild or stainless steels and the emissivities of the surfaces may vary from 0.16 for polished stainless steel to 0.81 for highly oxidized steel”.

Heat transfer through mould and lay-up

The following is a summary of the heat transfer mechanisms through the mould and lay-up (see Figure 3):

1) Conduction heat transfer through the diaphragms, composite prepreg, air gaps and mould materials.

2) Heat transfer across contact resistances between the various plies within the lay-up.

3) Natural convection within the enclosure between the inside of the mould and the bottom diaphragm.

4) Radiation across the air gap between the inside of the mould and the bottom diaphragm.

Conduction heat transfer through the diaphragms, mould and air gap may be solved analytically. Conduction through the prepreg is made complex by the rapidly changing state of the prepreg as it melts. The thermal properties of the material also change quite significantly. The specific heat of APC-2 increases from 925 J kg-’ K-’ at 50°C to 1670 J kg-’ K-’ at 380”C2. The contact resistance across the plies is pressure sensitive and will also change during heating as the prepreg softens and melts. The importance of radiation heat transfer across the air gap will depend on the size of the temperature difference and the emissivities of the mould and diaphragm surfaces.

EXPERIMENTAL DESCRIPTION

Description of UCG experimental autoclave

The experimental autoclave used for this work was custom designed and assembled in the Mechanical Engineering Department at UCG’. The autoclave (see Figure 4) consists of a pressure vessel designed and constructed to BS 5500 standards with a maximum design pressure of 2.06 MPa (20.6 bar). The heating system consists of 3-band heaters, with a total power output of 36 kW, which cover the entire outer surface of the central steel cylindrical sleeve. The vessel is insulated by wrapping insulation, 75 mm thick around the outer surface of the heater bands. Insulation is also packed into the end and door region.

A fan circulates the gas (air/nitrogen) through the ducts at the top and bottom of the vessel and back down the centre channel. The mean air velocity through the working section of the autoclave has been measured to be approximately 0.5 m s- I. The cooling system consists of a water cooled heat exchanger consisting of four rows of 12.5 mm outer diameter tubes in a staggered arrangement with 56 tubes in total. The door opens horizontally and the moulds are supported on two rails bolted perpendicularly to the door. Pressurization is achieved by cylinders of compressed nitrogen gas and the moulds are supported on two rails bolted perpendicularly to the door.

All temperatures are recorded using type K thermo- couples which have a maximum operating temperature of 1300°C with an accuracy of + / -3°C according to

Composites Manufacturing No. 314 1991 235

I- 1.075m

-l Door rnermocoupue

Circulation fan AI I Air Air

1 thermocouple

Top duct Inner steel cylinder

Pressure vessel

Wall at side of heat exchanger w

Air . . . . thermocouple

Band heaters

Heat exchanger

Bottom duct

Figure 4 Diagram of the UCG experimental thermoforming autoclave

BS 104111. Two permanent thermocouples are fitted within the autoclave in order to control the autoclave air temperature. One thermocouple is attached to the autoclave inner side wall (see Figure 4) to record the side wall temperature. It is held in contact with the side wall by a thin metal strip which is welded to the wall. The second thermocouple is sheathed, and measures the air temperature (Figure 4 ) .

An IBM PC is used to carry out the data acquisition and the PID air temperature control during processing. The temperature and pressure measurements are sampled every 20 seconds and written to a data file every 60 seconds. In order to reach the required air processing temperature quickly, the control system allows the side wall temperature to overshoot its steady state tempera- ture. As the air approaches the required temperature, the control system switches the heaters off and on for calculated time periods until steady state conditions exist. During the processing of APC-2 (38O”C), the side wall temperature may reach 450°C and is usually between 400°C and 410°C for steady state conditions during the APc-2 process cycle.

Jple

Experimental approach

The experimental work is carried out in two stages to estimate :

1) Heat transfer and energy flows to a mould and lay-up within an autoclave.

2) Heat transfer and thermal resistances through the typical mould and lay-up materials.

The first set of experiments attempts to examine the heat transfer to the top surface of a mould and lay-up within an autoclave. Since the top surface of a lay-up is flat during heating it was decided to replace the mould and lay-up by a flat metal plate insulated on the bottom side to act as the top surface of the lay-up. By changing the surface properties of the plate the convection and radiation heat transfer to the top surface of the lay-up are estimated. For the second set of experiments it was decided to measure the temperature distribution through the mould and lay-up, the air temperature and the wall temperature, to determine where the major resistances to heat transfer exist during the process.

236 Composites Manufacturing No. 3/4 1991

Heat transfer experiments for flat plates in the autoclave

The experiments estimate the relative importance of radiative and convective heat transfer to a flat plate within the experimental autoclave. The convection heat transfer is estimated first by placing a plate with a low emissivity surface in the autoclave and recording the heat transfer results. A plate with a high emissivity surface and the same dimensions and thermal properties as the low emissivity plate is then heated and, knowing the convection heat transfer from the first step the radiation heat transfer within the autoclave is estimated. The results are compared directly using the heat-up curves for the two plates. The steps in the experimental approach are :

Step 1. The convection coefficient (h) is estimated for a flat plate at different positions on the centre-line of the autoclave by using the heat-up results for a low emissivity (polished plate) within a non-preheated autoclave.

0.636 g and 380 J kg-l K-l respectively. The plate temperature is measured by three thermocouples placed equidistantly across the width of the plate (see Figure 5). The thermocouples are placed between the bottom surface of the plate and the insulation with the weight of the plate ensuring good contact between the thermocouples and the plate surface. Since the Biot No. of the plate is less than 0.1, the temperature distribution through the thickness of the plate is assumed uniform” (Biot No. = 1.4 x 10e4, based on h = 10 W mm2 K-l, L, = 1.5 mm and k = 110 W m-l K-l)“. The door, side wall and air temperatures are measured and recorded during all experiments.

Step 2. The reliability of the experimentally determined convection coefficients are confirmed by compar- ing these results to the heat-up results for a low emissivity plate within a preheated autoclave via a simple model (described below ) .

Step 3. Estimate the increase in heat transfer due to radiation in (i) a non-preheated and (ii) a preheated autoclave by recording the heat-up results for a high emissivity (black) plate and directly comparing experimental results for the low emissivity and high emissivity plates.

Step 4. Model the heat transfer to both the low emissivity and high emissivity plates and compare experimental and model results to examine whether the estimated convection coefficients and radiation model are realistic.

Heat transfer through a typical mould and lay-up

Diaphragm forming of complex curvature parts has proved successful’,’ and various experimental moulds have been designed at UCG’. These range from small simple flat plates to 1 m wing sections. A female hemispherical mould similar to that in Figure 1 (a) is used for the tests described below. The experimental approach to studying the important issues in heat transfer through this mould and lay-up is based on two objectives. These are:

1) To estimate the relative importance of the different resistances to heat transfer within the mould materials

Side wall

Duct

The convection coefficients can be determined from the following considerations. An energy balance on a flat plate being heated in an autoclave gives :

Insulation (15 mm thick)

4 tot = %onv + &ad (1)

For the polished plate used in Steps 1 and 2, qrad is assumed to equal zero. Making appropriate assumptions, Equation ( 1) becomes,

mc, dT,,/dt = hA,( T,+ - T,)

Therefore, h is estimated from (2)

+ Support rails

h = (mcpdTpldt)l(-&(T,ir - Tp)) (3) where ( Tair - T,) and dT,/dt are determined from the experimental results of Step 1. Step 2 is used to check the accuracy of Equation (2) and the calculated convection coefficients from Step 1 by using a completely different process cycle and set of experimental data. For the high emissivity plate used in Step 3, qrad in Equation (1) is not equal to zero and cannot be ignored.

/

Experimental plate

l.Om

positions

The experimental plates (see Figure 5 ) are made from brass and are 350 mm wide x 140 mm long and 1.5 mm thick. The width was chosen so the plate could be supported on the existing support rails of the autoclave. The original surface of the brass is highly polished and provides the low emissivity surface for the first plate. The emissivity of brass ranges from 0.028 for a highly polished surface to 0.09 for a polished surface”. The high emissivity surface is obtained by treating the surface with a thin layer of carbon (lampsoot ) which has an emissivity of 0.951°. The mass and specific heat of the plates are

1 /G Door

Figure 5 Schematic diagram of size and typical position of experimental plate in relation to the autoclave interior and support rails

Composites Manufacturing No. 3 /4 1991 237

Thermocouple at top of lay-w

Clamping ring

/

Thermocouple at bottom of lay-up

Figure 6 Exploded view showing position of thermocouples used to measure the temperature distribution through the mould and lay-up in the hemispherical mould

and to compare these to the heat transfer resistances on the outside of the mould and lay-up. This is achieved by recording the temperature against time for the different layers of a mould and typical lay-up within a non-preheated autoclave.

2) To demonstrate the relative importance of contact resistances between the plies by comparing the temperature drop through the unconsolidated lay-up of objective 1 with the temperature drop recorded in a vacuum consolidated lay-up undergoing a similar heat-up cycle.

The two diaphragms are of Upilex R material and 24 plies of APc-2, with a (0”/90”),, lay-up. A relatively large number of plies (24) were chosen to give a more easily measurable temperature difference between the top and bottom of the lay-up during heating. The (P/90’) lay-up was chosen because APC-2 is a continuous unidirectional composite and the thermal contact resistance between the plies is probably greater than if the carbon fibres in the different plies were running parallel to each other. The temperatures at the top and bottom surfaces of the lay-up are measured (see Figure 6). Attempts at measuring the temperature of the inside and outside mould surfaces at high temperatures by Monaghan and Brogan I3 have proved to be inaccurate due to an unsatisfactory method of attaching the thermocouples at the high temperatures involved.

EXPERIMENTAL RESULTS

The heat-up rates (K min- ’ ) for the preheat experiments are a linear average over the first five minutes after the mould is placed in the autoclave. The heat transfer rate ( W ) results for the preheat experiments are based on this average heat-up rate. The heat-up and heat transfer rates for the non-preheat experiments are based on the lo-20 minute period of heating where the rise of plate temperature against time is close to linear. The set point temperature for all experiments is based on the processing temperature for APc-2 of 380°C.

Low and high emissivity plates

The main findings from the heat-up experiments of the two plates are:

1) The average convection coefficient over the plate shows little variation along the centre-line of the autoclave (Figure 7). The large variation in h found by Ghariban, Haji-Sheikh and Lou7 in their model autoclave is not apparent here. The average

:/‘- 55 155 295 435 575 715 855

Distance from door (mm)

Figure 7 Variation of the average convection coefficient over a flat plate with position along the centre-line ofthe experimental autoclave

400 400-

z " t P-

q Side wall temperature q Side wall temperature

+ Air temperature + Air temperature

A Low emissivity A Low emissivity

V High emissivity V High emissivity

0 0 1 1 I I I I I I I I 1 1 I I 0 0 5 5 10 10 15 15 20 20 25 25 30 30

Time (min)

Figure 8 Temperature against time for the low and high emissivity plates heated in a preheated autoclave at 575 mm from door

convection coefficient over the flat plate along the centre-line of the UCG autoclave is 10.23 W m-’ K-l.

2) The high emissivity plate heats up faster than the low emissivity plate during both preheat and non-preheat experimental runs. Figure 8 shows typical results for the low and high emissivity plates at x = 0.575 m. The heat transfer rate to the plates during preheat experiments, at x = 0.575 m, increases from 113 W for the polished plate to 270 W for the black plate (28 K min-’ to 67 K min- ’ ). The average heat transfer rate during non-preheat experiments, in- creases from 59 W to 83 W (15 K min-’ to 21 K min-‘).

238 Composites Manufacturing No. 3/4 1991

500 Mould and lay-up experimental results

s 350- e

300 - it 2 250- I $ 200-

f 150-

100-

50 -

0 Heater band temperature

+ Air temperature

a 55mm

V 575 mm

x 855mm

01 ' 4

I I I I I I 0 5 10 15 20 25 30

Time (min)

Figure 9 Temperature against time for high emissivity plate (E = 0.95), heated in a preheated autoclave, for three different distances from the door

500[ I

450

400 I

s 350- e

300 - t! 2 250- iT2 “a 200-

j 150-

100

50

0 0 5 10 15 20 25 30

Time (min)

Figure 10 Temperature distribution of mould and unconsolidated lay-up heated within a non-preheated autoclave

3) The black plate heat-up curves shows significant variations from the front (i.e. door end) to the back of the autoclave. Figure 9 shows that, for the high emissivity plate, the heat-up is significantly slower when the plate is near the autoclave door. It is also clear that when the plate is placed near the door, it is slower to reach the steady-state temperature than when it is placed far away from the door. The heat transfer rate for the preheat experiments, increases from 210 W at x = 0.055 m, to 270 W at x = 0.575 m (52 K min-’ to 67 K min-‘).

4) The polished plate heat up curves show little variation along the length of the autoclave. The average heat transfer rates are 115 W and 59 W (28.7 K min- ’ and 14.6 K min-‘) for the preheat and non-preheat experiments respectively.

It is also found experimentally that there is a symmetric temperature distribution across the width (350 mm) of the black plate with the outer edges at a higher temperature than the centre of the plate. This is probably caused by the increased radiation heat transfer to the outer edges of the plate which are closer to the side walls of the autoclave. The temperature difference is up to 25°C between the outer thermocouples and the centre thermocouple.

The results presented here are for the temperature distribution between the top and bottom of the plies. Figure 10 shows a typical temperature distribution through the mould and lay-up for the unconsolidated lay-up heated in a non-preheated autoclave. For the second experiment, the lay-up was consolidated using a vacuum pressure of 0.1 MPa (1 bar). The table below gives a summary of the average temperature difference through the lay-up for the unconsolidated (see Figure 10) and the vacuum consolidated (see Figure 11) non-preheat experiments. These average temperatures differences are based on the lo-20 minute period of heating.

Temperature difference between :

Heater band-Air temp (H-A)

Air temp.-Top of lay-up (A-T)

Top of lay-up-Bottom of lay-up (T-B)

Unconsolidated Consolidated W) (“Cl

80 79

57 71

42 16

The main findings from this analysis are :

1) The temperature drops for the unconsolidated sheet, H-A, A-T, T-B are all of the same order of magnitude.

2) The temperature difference across the lay-up de- creases dramatically from the unconsolidated lay-up to a consolidated lay-up. This can be observed by comparing Figures 10 and II.

A vacuum pressure of only 0.1 MPa (1 bar) was used to consolidate the lay-up. These results imply the following : 1) All thermal resistances between the heater bands and

the bottom of the lay-up must be studied if a realistic overall simulation is to be produced.

2) The contact thermal resistance between plies plays a significant role in the delay of the heat-up of the bottom plies.

Error analysis

An error analysis is carried out on the calculation of the convection coefficient using Equation (3 ) . The error is

500

400

300

200

100

0

0 Heater + Air temperature 0 Top of layup LI Bottom of layup

10 15 20 25 30

Time [min)

Figure 11 Temperature distribution of mould and vacuum consoli- dated (0.1 MPa) lay-up heated within a non-preheated autoclave

Composites Manufacturing No. 314 1991 239

calculated using addition in quadraturei4. The estimated errors are :

Plate temperature diff. (dTJ + / -2°C (based on 144°C)

Air-plate diff. CL, - Tp) +/-1OC (based on 113°C)

Mass of the plate (m) +/ -0.0001 kg (based on 0.636 kg) Plate area (A,) +/-4 x 10m6m2 (based on 0.049m’)

The plate temperature difference is based on the 10 minute period described previously during the non- preheat experiments using the polished brass plate. The error analysis results are for the worst possible case. The accuracy of type K thermocouples is + / -3°C at 400°C”. The error estimate for the air-plate difference ( Tai, - T,) includes any error in the plate temperature due to radiation heat transfer to the polished plate. This error estimate is obtained from the model results assuming the plate has an emissivity of 0.05. The error is calculated to be 9% for the worst possible case and the average error is 8.5%.

MATHEMATICAL MODEL

This section describes the mathematical model for the heat transfer to a flat plate within the autoclave as described in the experimental method. The energy balance at the plate is as defined above,

4tot = 4.X”” + &ad (1)

where,

4tot = mcp dT/dt (4)

Forced convection heat transfer to the flat plate

The forced convection heat transfer rate to the plate is,

4 cow = hAp ( L - Tp ) (5) The convection coefficients (h) over the flat plate could

be estimated using correlations such as those found in Incropera and Dewitt . I2 This approach is not feasible within the autoclave as the air velocity over the plate at different locations would have to be known and is beyond the scope of this paper. The convection coefficients used in the model described here are determined experiment- ally from the low emissivity plate experiments as described in the experimental approach.

Radiation heat transfer to the flat plate

The radiation heat transfer may be solved analytically. The net radiation heat transfer, qi, from a surface, i, in an N surface enclosure is given by”

aT” - Ji 4i = 1 - Ei/EiAi = jg ,;;;-1 (6)

I 1 The experimental autoclave has six radiating surfaces consisting of :

1) Left side wall. 2) Right side wall. 3) Top duct plate. 4) Bottom duct plate. 5) Heat exchanger. 6) Door.

Figure 12 shows a network representation of Equation (6) for the heat transfer between the plate, duct wall and the two side walls within the autoclave. To model the heat transfer to the plate at any position along the centre-line of the autoclave using Equation (6), a completely defined enclosure must be specified. Since the bottom surface of the plate is insulated only the surfaces in the top half of the autoclave radiates directly to the top surface of the plate. However, the surfaces in the bottom half of the autoclave cannot be ignored as there is radiation exchange between them and the top surfaces of the autoclave. Hypothetical surfaces12, along the centre-plane of the autoclave must be defined as in Figure 13(a) to complete the enclosure necessary to use Equation (6). There are now ten surfaces (N = 10) within the enclosure.

b UT4 Pl

Figure 12 Network representation of the radiation heat transfer between some of the interior surfaces of the autoclave

Heat exchanger Top duct plate hypothetical surface I n

Bottom duct

Hypothet&l surface ‘Experimental

a defined on centre plane plate

Duct wall

b

Side wall

E3

Hypothetical surface for side wall

Plate

Figure 13 (a) The ten surface enclosure including the hypothetical surfaces defined on the centre plane and for side walls necessary for the radiation heat transfer analysis within the top half of the autoclave; (b) front view of autoclave showing hypothetical surface for side wall in autoclave radiation analysis

240 Composites Manufacturing No. 3/4 1991

The temperature of the plate is determined at each time step by rewriting Equation (4) as follows,

J-It+ 1 _ qtot At + g-n -~ P

mCP P

where,

t=nAt and the integer n is the number of time steps.

The radiation and convection heat transfer at the (n + 1) th time step are solved using the temperatures at the (n)‘h time step. The time step (At) used is 60 seconds. The convection heat transfer is calculated using the experimentally measured convection coefficient. In order to solve the radiation heat transfer, the radiosities (J) for the different surfaces must be determined. Equation (6) is written for the ten surfaces within the enclosure in matrix form ;

and {J} is determined by solving, (9)

The known temperatures in Equation (10) are at the (n)“’ time step. The radiation heat transfer to the plate is then calculated as follows,

qrad = . tsT; - Jp

I 1 1 - Ep/ &pAp

The view factors (Fij) between the different surfaces are determined using correlations developed by Chekhovskii et ~1.‘~ for parallel and perpendicular plates of arbitrary sizes. To calculate the view factors to the curved side walls using correlationsi another hypo- thetical surface as in Figure 13(b) is defined.

The side wall, door and air temperatures recorded during the experimental tests are used in the model. The duct plate and heat exchanger temperatures are assumed to be the same as the air temperature. The temperatures of the hypothetical surfaces on the centre plane are also assumed to be the same as the air temperature as they act as the hypothetical surface for the bottom duct plate. The time step (At) used in the model is 60 seconds because the temperatures are recorded by the data acquisition system every 60 seconds.

MODEL RESULTS

The emissivities for the polished and black plates are assumed to be 0.05 and 0.951° respectively. The emissivities of all the interior surfaces are assumed to be 0.81”. The interior surfaces including the heat exchanger are made from mild steel and have heavily oxidized surfaces (rust) due to long-term use. Oxidized steels have high emissivity values (0.66-0.81)‘O and are therefore very efficient at emitting radiant heat. The calculated convection coefficient at the different positions along the centre-line of the autoclave are also used in the model. The model reads in the temperatures of the side walls, door and air from the experimental data file. The heat exchanger, duct plate and hypothetical surfaces along the centre-plane are assumed to have the same temperatures as the air. Measurement of the duct plate temperature shows the duct plate to be very close to the air

s e 300 -

f! 3 z $ 200- E

E

100 t

0 L

ir temperature

A Experimental

X Model

0 5 10 15 20 25 30

Time (min)

Figure 14 Comparison of model and experimental results for the polished brass plate, at 575 mm from door, heated in a non-preheated autoclave. The emissivity of the plate is assumed to be 0.05

s e 300 -

f a c $ 200- E

F

a Side wall temperature

+ Air temperature

A Experimental

V Model

J 0 2 4 6

Time (min)

8 10

Figure 15 Comparison of model and experimental results for high emissivity plate (E = 0.95) at 575 mm from door, heated in a preheated autoclave

temperature and the air circulates through the heat exchanger.

Figure 14 shows the results using the model, with E = 0.05 for the polished plate in a non-preheated autoclave. The net heat transfer to the plate is over predicted by 6.8%. The percentage differences in heat transfer are based on the experimental results for the lo-20 minute period on the heat-up curves.

Figure 15 shows the results of the model for the black plate at x = 0.575 m (E = 0.95) in a preheated autoclave. The maximum temperature difference between the experimental and model results in Figure 15 is 21 K. The model overestimates the heat transfer rate to the plate by 6.4%. The model fails to predict the heat-up curve of the black plate at the position closest to the door during a preheat experiment and over estimates the heat transfer rate close to the door by 31%.

DISCUSSION

The lower mean air velocity within the working section of the UCG autoclave or the complexity of the air flow peculiar to different autoclaves may explain the small

Composites Manufacturing No. 314 1991 241

variation in the convection coefficient along the length of the UCG autoclave compared to the results of Ghariban et al.‘. Radiation heat transfer within the UCG autoclave may be somewhat higher than in other autoclaves due to the direct heating of the side walls and the location of the heaters. However, it is clear that the radiation heat transfer to moulds within all autoclaves is a very important mode of heat transfer especially when the autoclave is preheated.

The model results indicate that the radiation heat transfer may be modelled satisfactorily using analytical methods. The model developed here is for the heat transfer to a simple flat plate heated in the UCG autoclave. The average convection coefficient to this plate may be determined relatively easily by direct measure- ment. As the mould geometries become more complex the determination or prediction of the convection coefficient will become increasingly more difficult. The cold door reduces the heat transfer to a mould close to it during preheat cycles and methods of counteracting this reduction in heat transfer will need to be investigated in future.

CONCLUSIONS

1)

2)

3)

4)

5)

The heat transfer convection coefficient does not display great variations along the length of the UCG autoclave. The radiation heat transfer from the interior surfaces within the experimental autoclave is high (between 45%-57% of the total heat transfer to the plate with E = 0.95 during the preheat cycle and an average of 29% of the total heat transfer for the non-preheat cycle ) . Although the analysis carried out in this paper deals with one experimental autoclave, the above results suggest that radiation heat transfer within all autoclaves and to moulds in general may be important. The external resistance and internal resistances to heat transfer for a mould and lay-up are of similar orders of magnitude. Contact resistances between plies of a lay-up cause significant delays in material heat-up.

The results presented are for the UCG autoclave which is not of conventional design8*9. Some general comments are therefore appropriate. For autoclaves such as References 8 and 9, the radiant contribution to heat transfer would normally tend to be lower. When techniques are used to enhance the outside heat transfer to lay-ups and moulds (e.g. higher speed and more turbulent air flows, infra-red mould heating etc.), then the thermal resistances inside the mould/lay-up (e.g. contact resistances) will play a more dominant role in determining the heat-up time. Continuing research includes an extension of this heat transfer to other autoclaves and other techniques for thermoplastic processing.

ACKNOWLEDGEMENTS

This project was financially assisted by the BRITE program of the Commission of European Communities, Brussels. We would like to thank Professor P. Mallon, Mr M. Monaghan of University of Limerick who designed the autoclave at UCG and the technical staff at UCG’s Department of Mechanical Engineering.

REFERENCES

1

2

3

4

5

6

I

8

9

10

11

12

13

14

15

Monaghan, M.R. and Mallon, P.J. ‘Development of a computer controlled autoclave for forming thermoplastic composites’ Composites Manufacturing 1 No 1 (1990) pp 8-14 Murphy, P.F. ‘Mould design for advanced fibre reinforced thermoplastic composites’ MEngSc Thesis (Department of Mechanical Engineering, University College Galway, 1990) Guceri, S.I. ‘Transport phenomena in composites manufacturing- research issues and opportunities’ Proc XXII Znt Symp on Manufacturing and Materials Processing, ZCHMT, Dubrovnik, August I990 (Hemisphere Publishing Corporation, Washington, DC, USA, July 1991) Ogale, A.A. and McCullough, R.L. ‘Physical aging characteristics of polyetheretherketone’ Comp Sci Tech 30 (1987) pp 137-148 Ogale, A.A. and McCullough, R.L. ‘Influence of microstructure on elastic and viscoelastic properties of polyetheretherketone’ Comp Sci Tech 30 (1987) pp 185-202 O’Bradaiuh, C.M. and Mallon, P.J. ‘Effect of forming temperature on the properties of polymeric diaphragm formed-components’ ASCMICCM Joint SvmDosium on Composite Materials Science and Engineering, lJniver.&y of Delaware, USA, 23-25 September 1987 Gharihan, N.S., Haji-Singh, A. and Lou, D.Y.S. ‘Heat transfer in autoclaves’ Proc XXII Znt Symp on Manufacturing and Materials Processing, ZCHMT, Dubrovnik, August 1990 (Hemisphere Publishing Corporation, Washington, DC, USA, July 1991) SCHOLZ Brochure Ii 102/88 (Maschinenbau Scholz GmbH + Co. KG., Postfach 1649 (P.O. Box 1649), 4420 Coesfeld i.W., Germany) LBBC Hot Almosphere Autoclaves Brochure (Leeds and Bradford Boiler Company Limited, Stanningley Pudsey, West Yorkshire, England) Siegel, R. and Howell, J.R. Thermal Radiation Heat Transfer (McGraw-Hill, New York, 1981) BS 1041 Code for Temperature Measurement: Parr 4 Thermo- couples (British Standards Institute, London, 1966) Incropera, F.P. and Dewitt, D.P. Fundamentals of Heat and Mass Transfer Second Edn (John Wiley & Sons, New York, 1985) Monaghan, P.F. and Brogan, M.T. ‘An overview of heat transfer for processing thermoplastic composites in autoclaves’ Proc Flow Processes in Composites Materials Conference, University of Limerick, Ireland, 4-5 July 1991 Collett, C.V. and Hope, A.D. Engineering Measurements Second Edn (Pitman, 1983) Chekhovskii, I.R., Sirotkin, V.V., Chu-Dun-Chu, Yu.V. and Chebanov, V.A. ‘Determination of radiative view factors for rectangles of different sizes’ (translation of Russian original) High Temperature 17 No 1 (1979) pp 97-100

AUTHORS

P.F. Monaghan is in the Department of Mechanical Engineering, University College Galway, Ireland. M.T. Brogan is with the Thermal Engineering Research Unit (CATERU), Manufacturing Research Centre, Univer- sity College Galway, Ireland. P.H. Oosthuizen is with the Department of Mechanical Engineering, Queen’s University, Kingston, Ontario K7L 3N6, Canada. Correspondence should be addressed to P.F. Monaghan.

242 Composites Manufacturing No. 314 1991