Heat Transfer in The Furnace of Shell Type Boiler [PDF]

Zitiervorschau

809

STEAM PLANT GROUP

HEAT TRANSFER IN THE FURNACES OF SHELL TYPE BOILERS D. C. Gum* Peak heat fluxes which occur near the oil burner can be predicted by a simple mathematical model to within 12 per cent and thus help to estimate peak metal temperatures. Although peaks of 150 000 Btu/ft2 h can occur in current practice, the metal of the furnace is not at risk provided the waterside is clean. Conversely, the risk is great when even a small amount of scale has built up.

1 MODERN SHELL BOILERS

WITHINthe last 20 years shell boilers have changed from large Lancashire and Economic type plant, mostly coal fired and manually controlled, to fully automatic units of much smaller dimensions, fired with oil or gas. Surprisingly, this has been achieved with no increase in capital cost per unit of heat produced and the thermal efficiency has been raised from around 70 to over 80 per cent of the gross calorific value. A typical shell boiler of British design is shown in Fig. 1. The reversal chamber at the rear end is entirely surrounded by water in this 'wet-back' type; in the 'dryback' the rear part is lined with refractory. The essential purposes of a boiler furnace are to accommodate the combustion process and to reduce the temperature of the combustion products before they reach the tube plate in the reversal chamber to an acceptable level, say 1750°F. In the larger snell boilers furnaces are commonly about 4 ft in diameter and 12 ft long, the metal being about $ in thick. For up to about 35 000 lb/h evaporation one furnace is used, for larger outputs two may be required. 1.1 Notation A Heat receiving area, ft2. Specific heat at constant pressure, Btu/lb degF. c, D Furnace internal diameter, ft. d Radiation beam length, ft. F Heat flux, Btu/ft2 h. G Mass flow rate of furnace gases, lb/ft2 h. Net heat input to furnace, Btu/h. H J Difference between temperatures of scale and water at interface, degF. K, Thermal conductivity of metal, Btu/ft2 h degF in. K , Thermal conductivity of scale, Btu/ft2 h degF in. k Shape factor of carbon particles in flame. L, Metal thickness, in. L, Scale thickness, in. This paper is published for written discussion. The M S . was received on 12th January 1973 and accepted for publication on 18th October 1973. 21 * Research Director, Thompson Cochran Division, Clarke ChapmanJohn Thompson Limited.

M n

Million(s). Number of carbon particles in flame. Qc Convective heat transfer, Btu/h. QE Heat leaving a defined zone in the furnace, Btu/h. QR Radiant heat transfer, Btu/h. QT The sum of radiant and convective heat transfer, Btu/h. T , Absolute flame temperature, OR. T, Absolute sink temperature, "R. T,, Mean metal temperature above water temperature, degF. r, Effective flame temperature, O F . t, Ambient temperature, Temperature difference across metal, degF. t, Temperature difference across scale, degF. rs V Furnace volume, ft3. Index of 'D' in expression for emissivity. y E Emissivity. 0 Stefan-Boltzmann constant, 1-72x lo-" Btu/ft2 h (degR)4. O F .

2 FURNACE HEAT ABSORPTION

A series of heat transfer tests have been carried out on modern shell boilers, using both gas and heavy oil in different types of dual-fuel burners (I)?. By measuring the temperature of the combustion products as they left the furnace by a suction pyrometer, as well as their composition, the enthalpy at this point could be obtained. When deducted from the heat input this yielded the furnace heat transfer. Fig. 2 shows that for both gas and heavy oil firing, the relationship between heat input and the percentage transferred in the furnace appears to be .linear. Per cent heat transfer decreased with increasing load and was considerably less for gas than for oil because the gas flame was almost clear and did not emit the carbon radiation which was evident with the heavy oil. A typical modern boiler was used. The furnace heat transfer area was only 6.5 per cent of the total boiler heating surface and yet, with oil, it transferred 40-50 per cent of the total heat input; with gas, 30-40 per cent. There is some scatter in the points, especially with gas

t

References are given in the Appendix.

Proc lnstn Mech Engrs 1973

VOI 187 64/73

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

D. C. GUNN

810 Transfer box

Water l e v e l

‘\

\

\ I

C h I m ney

I‘

\

&

i

Burner __ +

iC o m b u s t ion chamber

Fig. 1. Diagram of contemporary wet-back shell boiler

The furnace area of the boiler tested was 120 ft2 and the maximum gross heat input was 19.66 MBtu/h on oil. At this loading the heat transfer to the furnace was 40.8 per cent of the input. The average heat flux in the furnace was therefore 67 000 Btu/ft2 h. With gas the maximum input was 21.96 MBtu/h of which 27.3 per cent was transferred, equivalent to an average heat flux of 50 000 Btu/ft2 h. In spite of the higher temperatures of combustion products with gas firing, the tube-plate metal temperature was not significantly higher than with oil; this was attributed to some residual carbon radiation from the gases in the reversal chamber when working on oil. The tube-end temperatures were, however, higher, especially in a dry-back boiler which was subsequently tested. 3 PATTERN O F HEAT TRANSFER

It can be expected that the longitudinal pattern of heat

10

12

14 16 Heat input

18

20

flux is non-uniform, with a peak that may overheat the furnace metal. Some 20 years ago the Instituut voor Warmte-Economie (T.N.O.) Delft evaluated the metal temperatures in the furnaces of a Cornish boiler with a single corrugated furnace, 24.6 ft long. The furnace diameter was 3.3 ft and its heated surface was 285 ft2. The operating pressure was 114 lbf/in2. Thermocouples measured the metal temperatures on the water-side and the gas-side at various points along the furnace at various loads, using different types of oil burner at varying rates of heat input (3). Fig. 3 shows the maximum metal temperatures against furnace length for two of the oil burners; burner ‘B’ was regarded as unsatisfactory and was later modified by the manufacturers to give lower peak values. Fig. 4 shows the heat fluxes, calculated from the difference between gas-side and water-side metal temperatures. The firing rate for ‘A’ was 2.2 MBtu/ft h and for ‘B’ 1.78 MBtu/ft h, diameter being included in the parameter. The peak metal temperature with burner ‘B’ was 610°F and 517°F for burner ‘A’ in spite of the lower firing rate with burner ‘By. In our tests, referred to previously, the furnace diameter was 3.5 ft and the maximum thermal input was 2344 MBtu/h, giving 6.72 MBtu/ft h. This was very much higher

106 B t u / h

Fig. 2. Furnace heat transfer and heat input, w i t h CO, noted

firing, which could have been caused by non-uniform gas temperature at the furnace exit and also by CO, variations. With oil the CO, range was 11.8-12.3 per cent and with gas 8.7-9.2 per cent. The effects of air dilution on flame radiation are likely to be more marked with gas than with oil, since non-luminous emission from CO, and H,O is the only source of radiation, whereas these effects with oil are likely to be swamped by carbon radiation. The C02 concentrations are noted in Fig. 2 and, with gas firing, the points of higher concentration lie above the line. With gas the heat absorption in the furnace might be greatly improved by working nearer to the stoichiometric point (11.5 per cent CO,). Most commercial burners can not yet do this reliably with acceptable levels of CO (2).

I 400

L

, I

0

2

L

I

4 Distance f r o m b u r n e r

6

ft

Fig. 3. Furnace metal temperatures w i t h t w o oil burners, from T.N.O. Information 29 Vol 187 64/73

Proc lnstn Mech Engrs 1973

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

HEAT TRANSFER IN THE FURNACES O F SHELL TYPE BOILERS

811

*Or

40~

70 60 -

30 -

Burner A

20 Burner 0 I

I

01 0

2

I

6 Furnace length

4

I

I

8

10

ft

Fig. 4. Heat flux patterns obtained on an oil-fired Cornish boiler

700-

PI L + 3

,?

600-

W

a E

.W

- L L

2.0 E

5

.-5 X

0

z

400

0

5

Fig. 6. Permissible net heat inputs t o boiler furnaces, T.U.V. (Germany)

/Burner A

B u r n e r B, m o d \ f led

1

4

3 ft

Burner B, o r i g i n a l

/

500-

I

2

Furnace diameter

k;. I

lot

2 3 F i r i n g r a t e HID M Btu/ft h

4

5

Fig. 5. Peak metal temperatures in an oil-fired Cornish boiler

than in the Dutch tests, showing the progress made during the last 20 years. Nor is this the highest firing rate now used in shell boilers. Fig. 5 shows how maximum metal temperature increased with firing rate. Burner ‘A’ represents the type used in British boilers with this order of furnace size. The relationship over the range of the data is linear and extrapolation suggests that some very high temperatures could be reached. The graph also shows burner ‘Byas originally designed, and as modified by flaring out the quarl more sharply. The original line rises steeply; as modified, the line is nearly parallel to, but below, the line for burner ‘A‘. This shows the importance of burner design. 4 LIMITATION TO HEAT INPUT

More recent work by the T.N.O. (4) has suggested certain limitations to the firing rate in shell boilers since the

maximum metal temperature must be below 750°F with common boiler steels, or 842°F for 15. Mo. 3 alloy steel. Also, the maximum heat flux anywhere in the furnace shell must not exceed 92 000 Btu/ft2 h. I n West Germany the T.U.V. (Technischer uberwachungs-Verein) (5) limits the thermal input according to the diameter of the furnace. Fig. 6 illustrates this standard: for non-alloy steel the heat input to a furnace reaches an absolute limit of 27 MBtu/h at a furnace diameter of 140 cm (55 in); even with larger furnace diameters the heat input must not increase. With alloy steels the limit can be raised to 36 MBtu/h at the same furnace diameter. There could be theoretical justification for this rule on the grounds of increased flame radiation with increased furnace diameter. I n the U.S.A. the Federal Construction Council (6) recommends that the gross heat input per unit area of effective radiant heated surface should not exceed 100 000 Btu/ft2 h and the volumetric gross heat input should not exceed 150 000 Btu/ft3 h. Thus, in several countries there are either recommendations or mandates limiting the heat inputs to shell boiler furnaces, presumably as a result of operating experience of furnace distortion (3) or high exit temperature (6). The subject seems to be of sufficient importance for further inquiries to be made. 5 FURNACE HEAT FLUX

Site conditions prevented the attachment of thermocouples in the hot zone of the furnace in our own tests: at the exit end the highest temperature was 599°F on oil and 565°F on gas; these are not high enough to cause concern. Other experiments in open tanks were made on a furnace of 2.75 ft diameter with a gross heat Vol 187 64/73

Proc lnstn Mech Engrs 1973

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

D. C. GUNN

812

Table 1. Furnace metal temperature for a heat flux of 100 000 Btu/ftz h

Boiler pressure, Ib/in2

Furnace thickness, inches

11 i

1

"F

Without scale

I 1

0.375

100 200 300

Furnace fireside temperature,

0.625

0,875

486 615

729

I

1

With 0.02 in scale 614 743 857

input rate (on oil) of 19-76MBtu/h, giving HID = 7.2 MBtu/ft h and H/V = 355 000 Btu/ft3 h, both high figures. The metal temperature recorded 2 ft from the furnace front was 398°F with free boiling. The radial distance between thermocouples was 0.455 in and the heat flux, calculated on a basis of a water temperature of 212°F and a temperature drop of 32 degF through the waterside film was 107 000 Btu/ft2 h. Even 100 000 Btu/ft2 h could be severe enough with a thick furnace, as is shown in Table 1. I n boilers at 200 lbf/in2 with only 0.02 in scale, the Dutch limit of 750°F is approached. The answer seems to be that no scale must be permitted to form, the problem obviously becoming greater at higher heat fluxes and pressures. Another series of measurements in a smaller tank with a furnace 1.93 ft diameter showed maximum temperatures 2.25 ft from the furnace front. With a gross heat input of 4-74 MBtu/h (HID = 2.46 MBtulft h), the peak was 360°F in a freely boiling tank, equivalent to a heat flux of 65 500 Btu/ft2 h. In collaboration with a burner manufacturer (7) a large rotary cup burner was fitted to a furnace with an internal diameter of 3.75 ft. Thermocouples were fitted into the 0.875 in thick crown, 0.503 in apart radially, with pairs at intervals of 6 in. This device formed a heat-flow meter. The tests were carried out at $, 3, t and full load (41.1 MBtu/h net heat input). A peak of 122 000 Btu/ft2 h

'*OI

TIwill not be the adiabatic temperature of the flame which

Key

x

Burner B, original Burner B. modified + Burner A A Author's early tank t e s t x Dr Wu o

/ /

o /A

/+ 40

2 . L l - L . U

2

4

6 MODELEVALUATESPEAKHEATFLUX

The heat transfer in a boiler furnace involves radiation and convection. The radiant heat transfer is governed by the Stefan-Boltzmann Law: QR = mA(TI4-Tz4) Since the flame temperature will be in the order of 3000°F and the furnace metal temperature 750"F, the latter can be neglected in the fourth-power relationship which can now be written as:

QR = acATI4

/

loo/

was measured at Station 4. Fig. 7 shows peak heat flux against heat input per unit furnace diameter and also heat fluxes deduced from the Dutch data in Fig. 5. The peak heat flux in an oil-fired cylindrical furnace is thus a function of the heat input per unit furnace diameter. Nearly the whole range of heat inputs of modern boilers has been covered. It seems that peak heat fluxes varying between 50 000 Btu/ft2 h for small boilers, and 122 000 Btu/ft2 h for larger boilers, can be expected. Wu (8) at the University of Delft has compared the behaviour of light oil and natural gas in a tank furnace of 1.97 ft I.D. by dividing the tank into axial sections and measuring flow of heat from each. The heat-flux distribution followed a pattern similar to that shown in Fig. 5 for burner 'A'. The peak occurred one furnace diameter downstream from the burner. The peak heat flux at HID = 1.79 was 62 000 Btu/ft2 h and is shown in Fig. 7 to be slightly above the downward extension of the curve through our own tank tests. Wu's experiments with natural gas show that 'when the reaction in the flame is 80 per cent complete, the fraction of energy transferred from the oil flame is twice that from the gas flame'. Wu has also shown that there is a marked improvement of heat transfer when swirl in a gas flame is increased, which emphasizes the importance of burner design. Furnace diameter as such can be expected to be of importance through its relationship to the beam length for radiation, and through the length/diameter ratio in convection (9). A very simple mathematical model demonstrates the relationship between peak heat flux, heat input and furnace diameter with fuel oil firing. Gas has been ignored as oil is the more potent, and hence the more hazardous, as regards peak heat transfer fuel.

6 8 Heat input, H/D M Btu/h

10

I2

Fig. 7. Heat flux and heat input, from various sources

loses heat to the furnace wall. The heat-flux patterns in furnace tubes all show that the peak occurs roughly at one furnace diameter downstream from the burner, suggesting that the combined effects of temperature and emissivity reach a maximum at about this point. The assumption will now be made that the combustion reactions are substantially complete at that distance and that temperature, emissivity and convection conditions within it therefore are uniform! This is a very sweeping statement, as the development of flame, emissivity and temperature depend on time, and therefore on furnace length, and on air/fuel mixing patterns. But unless combustion progress and aerodynamic conditions can be specified and the laws stated, a rigorous approach does not seem possible. Furthermore burners vary greatly and their settings depend on commissioning staff. Therefore any Vol187 64/73

Proc lnstn Mech Engrs 1973

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

HEAT TRANSFER IN THE FURNACES OF SHELL TYPE BOILERS

assumption is bound to be broad and the above may not be too unrealistic. The heat reception area can now be defined as: A = nD2 so that the radiant heat transfer becomes : Q R = aeaD2Ti4 The further assumption is made that in the zone defined by diameter D and length d, plug. flow at a uniform temperature t transfers heat Qoto a sink temperature of 750°F. Using a simple expression (10)for the convection coefficient: Qc = 0 * 0 1 4 4 ~ , G ~ ' ~ D - ~ ' ~ ( t ~ - 7 5 0 ) ~ D ~ the total heat transfer Qt in the zone is QT = crcaD2T14+0~0144~,G0'8D-0'2 (t 1-75o)nD2 The heat leaving the zone, QE,is :

':I 20 0

The sum of all these heats can be equated to the net heat input, H , as follows: H=

QR+Qc+QE

= a ~ ~ D ' (+460)4+0*0144D2~pD tl -O"(tl-

750)GO"

813

0

2

4 H e a t input,

6

8

10

HITTD'

10' ,Btu/ft* h Fig. 8. Calculated curves for peak heat flux and thermal input, H / d P . for furnaces 2, 3, 3.75. 4 and 5 ft diameter, with experimental points plotted for a furnace of 3.75 f t diameter Table 2. Calculated and measured peak heat fluxes

H -- ac(tl +460)4 XD2

G + 0 * 0 1 4 4 ~ ~ D ~ ' ~ ( t ~ - 7 5~0, )( G t, ~ta) ' ~ +(1) ~

It now remains to define emissivity E . For the present purpose it has been assumed that the non-luminous gas radiation is swamped by the carbon radiation and this must be related to D. The basic equation (XI) is: = l-e-nkd

3.75

The heat fluxes obtained in our tank experiments were used to obtain an emissivity which would make equation (1) fit the experimental results. This was 0-35 and enabled the constant in the exponent of e to be derived for the particular furnace diameter of 3-75 fr. This constanr embraces both n, the number of carbon particles, and the coefficient k. The exponent becomes 0.117D. Thus,

1.93

=

1-e-0.117D

. . . .

(2)

Using this value, equation (1) has been solved for total heat flux (QR+Qc) for furnaces of 2, 3, 3.75, 4 and 5 ft diameter and for values of H/nD2varying between 200 000 and 1 MBtu/ft2 h. It will be noted that the parameter H/rD2 is now used rather than HID, because it is conveniently grouped in equation (1). The solutions are plotted in Fig. 8 with the points obtained in the tank experiment inserted. The fit is reasonably good. Table 2 gives heat fluxes measured on the tank furnaces; the agreement between results derived from Fig. 8 and the measured values is within 10 per cent except for the furnace of 2-75 ft diameter where the model underestimates the measured heat flux at H/rrD2 = 425 000. Further experimental work should include an extension of the heat input rates to smaller furnaces and the

2.15

From Fig. 8

930 000 710 000 530 000 225 000

126 000 113 000 94 000 63 000

1 t;;::: 1

103 000 78000

1 1

1.97

Peak heat flux, Btu/ft2 h

H/rrDZ, Btu/ftz h

Furnace diameter, ft

1

333 000 273000 253 000 242000 289000

1

122 000 114 000 96 000 61 000

1

60 000 52 000

1

57000

Measured

107 000 88000 64 000 53 000

1

62000

exploration of the heat fluxes in 4 ft diameter and larger furnaces; particularly the latter, since Fig. 8 suggests that high heat fluxes could be encountered which would proportionally increase the metal temperatures shown in Table 1. The solution of equation (1) also gives the 'flame temperatures' occurring in the model furnace zone. This is the temperature needed to provide the corresponding heat fluxes with the 'emissivity' given by equation (2). The solutions obtained are shown in Fig. 9. A range of 2500-3300°F is indicated, the higher temperatures being associated with the smaller furnaces. At the upper end of the range dissociation of CO, accounts for some 5 per cent of the heat flux and is therefore significant. The exponent of e was estimated with H / r D 2 = 930 000 for a furnace diameter of 3-75 ft where the calculated temperature is above 3300°F. The measured temperature could be expected to fall short of this value but the

Proc lnstn Mech Engrs 1973

Vol 187 64/73

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

D.C. GUNN

814

diameter (4.6 ft) is 94 000 Btu/ft2 h for plain steel and 108 000 Btu/ft2 h for alloy steel.

33r

7 OTHER HEAT FLUX MEASUREMENTS

34-

II

31 I-

Heat input, HITI&

105 Btu/ft2 h

Fig. 9. Calculated flame temperature related t o heat input, H/rD2.f o r furnaces of 2 , 3 , 4 and 5 f t diameter

heat flux, estimated from Fig. 8, should be correct since the dissociation would be taken care of in a lowered value for E. In other words, the empirical value of E is below its true value. This would tend to give low estimates of heat flux at the lower end of the heat input range. The work of Wu (8) has indicated that this is indeed the case. Table 2 shows that Fig. 8 underestimates the heat flux by 9 per cent. Wu's measurement of emissivity in this case was 0.42, compared with 0.22 calculated from equation (2). It therefore seems that a better correlation would be obtained by modifying equation (2) to: = l-e-O-11'7DY where y is less than unity. It would be obtained by analysing results from tank tests with a wide range of diameters. . Wu's corresponding temperature for the heat flux, shown in Table 2, was 2370"F, compared with 3050°F from Fig. 9. The respective radiant heat fluxes are 45 500 and 56 000 Btu/ftz h, using the emissivities cited above. The balance between these and the values shown in Table 2 should be convective heat transfer, plus experimental error. The foregoing relates to furnaces of small diameter; it is the larger furnaces which are of greater interest to the boilermaker, since these generally operate with higher H/rrD2. Also, because of their increased diameters their thicknesses are greater so that they are likely to operate nearer the temperature limits of the metal. From Fig. 8 the probable heat flux can be read off. For a furnace 4.25 ft diameter and 16.4 ft long with a net input of 25.2 MBtu/h, the peak is 96 000 Btu/ft2 h, which compares with 82 000 Btu/ft2 h derived by the Dutch method. The peak heat flux from Fig. 8 which would relate to T.U.V. requirements for the maximum permitted furnace

Although it is difficult to compare coal, fired on a chain grate, with oil, burned concentrically in a furnace, similar peak heat fluxes have been obtained. Wright (12) has reported a heat flux of 100 000 Btu/ft2 h on the crown of a furnace, 3:5 ft in diameter, when fired with washed smalls at 11.4 MBtu/h. At this point heat flux was still rising steeply with firing rate. He also showed that the heat flux can be even higher at positions other than vertically above the centre line of the grate, reaching 150 000 Btu/ft2 h at 45". With the T.U.V. maximum permitted input of 18 MBtu/h for the 3.3 ft diameter furnace, the heat flux from Fig. 8 is 93 000 Btu/ft2 h for plain steel. This suggests that the T.U.V. regulation is based on a peak heat flux of about this value. For alloy steel, the permitted heat input for this furnace would be 23-9 MBtu/h and the corresponding heat flux 104 000 Btu/ft2 h. Thus, it seems that in two European countries maximum heat fluxes of about 100 000 Btu/ft2 h are mandatory. 8 CONVECTION

The convection component of heat transfer can be evaluated from equation (1). With oil-firing it appears to be of minor importance, varying between 1.7 per cent of the peak heat flux for a 5 ft diameter furnace with H/aD2 = 200 000 Btu/ft2 h up to 8.7 per cent for a 2 ft diameter furnace with H/rrD2 = 1000 000 Btu/ft2 h. With gasfiring, convection is of much greater significance than with oil. There has been no attempt made to estimate it from equation (1) but this could be done by using emissivities from the Hottel curves. The impingement of hot gases on the furnace wall would have an important effect, even with oil firing. A protective layer of carbon is often deposited but in certain types of flame hot gases, free from carbon, may impinge on the metal. Recent work (13) by the International Flame Research Foundation throws some light on the subject. A watercooled steel plate was placed 0.75 m in front of a gas burner in a refractory-lined furnace of rectangular cross section 2 m x 2 m. The maximum convective heat flux measured was 51 000 Btu/ft2 h with a gas temperature of 2550°F. The heat release rate was only 60 000 Btu/ft3 h, compared with about four times this figure in a shell boiler furnace, but local velocities may not have been very different. The actual angle of impingement appears to have been about 60".

Thus, if impingement occurred in the region of high radiant heat flux, the combined flux might reach 200 000 Btu/ft2 h which would be disastrous in a large furnace with water-side scale. This might explain why furnaces sometimes bulge in their lower parts. 9 IMPLICATIONS

In both Holland and West Germany there are rules limiting the heat input to furnaces to avoid overheating the metal. The Dutch rules alone provide guidance on how to estimate the metal temperature. I n the UK boilers are generally built to B.S.S. 2790 in the 1969 version of

Proc lnstn Mech Engrs 1973

Vol187 64 I73

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

HEAT TMNSFER IN THE FURNACES OF SHELL TYPE BOILERS

which there is a requirement (14)that, for furnaces, the mean metal temperature to be used for the evaluation of allowable stress shall be 90 K or 162 degF above the maximum temperature of the water. The mean metal temperature depends on heat flux and is not necessarily 162 degF above water temperature. Assuming the mean to be tt,; J+t,+$t,

400

815

I

/

= 162

J depends on heat flux, but is relatively small and assumed to be constant at 30 degF (15). LE! Ks

t, = F-

Lm Kin

t , = F-

and hence

T,, = 3 0 + F

”+-- 2) (is

Taking practical values :

L, = 0.02 in

L, = 0.875 in K , = 15.6 Btu/ft2 h degF in

0 40

K,,, = 312 Btu/ft2 h degF in the equation becomes :

T,, = 30+0.00366F

. . .

(3a)

. .

(3b)

60

80 100 Peak h e a t f l u x , F lo3 B t u / f t 2 h

120

140

Fig. 10

or, if no scale is present,

T,, = 30+0.001 3 8 F .

For metal 0.375 in thick the equation becomes respectively : T,, = 30+0.001 87F . . . ( 3 ~ ) and, T,, = 30+0-000 5 9 F . . . (3d) Fig. 10 illustrates these relationships and shows that for furnaces of all practical thicknesses and all practical peak heat fluxes, the B;S.S. 2790 mean temperature is realistic only in the absence of scale. With as little as 0.02 in (0.508 mm) scale the average metal temperatures exceed the B.S. figure, particularly for thick furnace walls and high outputs. Even without scale, the heat flux should be below 95 000 Btu/ft2 h. Therefore, when scale is present, the stress allowed by B.S.S. 2790 is too great and any factor of safety is being eroded as heat flux and scale thickness increase. The B.S.S. norm is not a maximum permitted temperature, it is the temperature at which the stress is calculated. If, as would seem logical, it was to be replaced by a figure dependent on heat flux, the effect would be to increase the permissible stress when no scale is present and to decrease it in the presence of scale. This emphasizes the need for water treatment. B.S. S . 2486, currently under revision, provides considerable guidance in such matters. The U.K. shell-boiler industry has prepared documentary help for boiler users (16).It would be interesting to study the types of furnace failures, not only in this country but also in those where limits are imposed on thermal input to see whether furnace failures are significantly reduced by such measures.

10 CONCLUSIONS

(1) Average furnace heat fluxes of around 70 000 Btu/ ft2hfor oil-firing and 50 000 Btu/ft2h with natural gas have been measured on a furnace of 3.5 ft diameter with an effective heating surface of 120 ft2 with around 20 MBtu/h gross heat input. (2) Experiments in Holland have shown that the heat flux in a furnace is non-uniform and reaches a peak about one furnace diameter downstream from the burner, the extent of the peak being strongly dependent on burner details. (3) More recent experiments on furnaces in this country indicate that, with ordinary burners using heavy fuel oil, peak heat fluxes can reach 122 000 Btu/ft2 h in a furnace 3.75 ft diameter and that lower heat fluxes occur in furnaces of smaller diameter. There is a need for further experiments to investigate more fully the effect of diameter. (4) In the absence of such experimental evidence, theory indicates that peak heat flux increases with firing rate and with furnace diameter, approaching 150 000 Btu/ ft2 h on a furnace of 5 ft diameter fired at a net heat input of 1 MBtu/ft2 h (H/n-D2).Within a range of furnace diameters 2-5 ft and HlrrD2 up to 1 MBtu/ft2 h the mathematical model used predicts the heat fluxes obtained on the limited experimental information available to within 12 per cent. (5) The associated theoretical flame temperatures vary between 3200°F for a 5 ft diameter furnace fired at 1 MBtu/ft2 h (H/rrD2)and 3300°F for a 2 ft diameter furnace fired at 0.6 MBtu/ft2 h, at which temperature dissociation becomes important. (6) Convection with oil firing is of minor importance,

Proc lnstn Mech Engrs 1973

Vol187 64/73

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

D.C. GUNN

816

but flame impingement could result in local heat fluxes of 50 000 Btu/ft2 h, in addition to the normal radiant flux. (7) Water-side scale is important. There is no risk to the furnace metal if the water-side is clean, even with the highest heat fluxes indicated in this paper. With even a thin layer of scale, the metal can overheat and distort. Correct water control is therefore essential: zero hardness should be the target. 11 ACKNOWLEDGEMENT

The author’s past colleague, Mr H. M. Simpson, C.Eng., M.I.Mech.E., was closely associated with the early work of this paper, and would have been a co-author had it not been for his retirement. APPENDIX REFERENCES

GORE,W. H.,GUNN,D. C., and HORSLER, A. G. Inst. Fuel National Meeting March 15 1972. Aston University. (2) HOGGARTH, M. L., POMFRET, K. F., and SPITTLE, P. ‘Evaluation of dual fuel burner performance.’ The Gas Council, Research Communication G.C. 196. (3) LUNING, 0. E., MULDER, L. L., and VERMEULEN, N. P. J. 1953 ‘Research made into the temperatures occurring in a

(I)

Cornish boiler.’ Information Report No. 29. Insr. ‘uoor Warmt-Econornie (T.N.O.) Delft, Netherlands. (4) Dienst Voor Het Stoomwezen: 1970 Doc. A.B.G. 28 (2) Den Haag: Netherlands. (5) T.U.V. Va T.U.V. Merkblatt Dampfkessel Richtlinien fur die Beurteiling von Kesselkonstruktionen. Dampfkessel 451-71/1 p. 13 Aug. 1971. ( 6 ) Boiler Rating Criteria for Non-Residential Boilers: 1962 Report No. 44 Publication 981: National Academy of Sciences: National Research Council, Washington D.C. (7) Hamworthy Engineering Limited, Poole, Dorset. ( 8 ) WU, H. L. ‘Comparison of the performance of natural gas and oil flames in a cylindrical furnace. J . Inst. Fuel 1969 (Aug) 42,417. (9) MCADAMS, W. H. Heat transmission 3rd Edition, Equation 9-17, 227. McGraw-Hill, New York. (10)Ibid. Equation 9-16, 226. (XI) Ibid. Equation 4-65, 100. (12) WRIGHT, S. J. Inst. Fuel 1965 38, 124. (13) International Flame Research Foundation A Digest of the NG-3 Trials Doc. G33/a/6, 1971. (14)B.S.S. 2790 Part 1 1969 Shell Boilers of Welded Construction 2.2.2. (15) B.S.S. 2486 Treatment of Water for Land Boilers (under revision) British Standards Institution, London. (16)Water Treatment for Shell Boilers. Association of Shell Boilermakers, Manchester.

Proc lnstn Mech Engrs 1973

Vol 187 64/73

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

Page No. D261-D262 (Discussion) missing

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

HEAT TRANSFER IN THE FURNACES OF SHELL TYPE BOILERS

maladjustment or deterioration of firing equipment can also aggravate the situation. It is interesting to recall that some measure of furnacetube protection was provided in the past by fusible plugs in furnace tubes but since these fell into disrepute because they failed to operate when encased in scale, no alternative safety provision has been introduced. It is to be hoped that boiler design engineers will continue to make some ‘scale allowance’ when deciding the maximum permissible heat flux rate for a given furnacetube diameter and, perhaps, boiler manufacturers will consider the provision of thermocouples attached to the furnace tube and/or rear tube plate in their packaged units. A suitable temperature indicator could also be included as a standard item and the system extended (particularly for partially attended boiler plants) to initiate lock-out of the firing system should an excess metal temperature be reached. The provision of metal-temperature indication should not, in any way, reduce the importance of correct. water treatment, but it would, at least, give an outside indication of what is happening within the boiler. The savings on costly repairs which result from damage due to overheating would, I feel sure, more than compensate the user for a marginal increase in the initial cost of the equipment. H. P. Heward Member The paper illustrates the high rates of heat transfer which are now being achieved in the furnaces of shell-type package boilers, and emphasizes the special care which such boilers require, particularly in the selection and accurate setting of burners and in ensuring proper feedwater treatment. I have felt for some time that with shell-type package boilers there has been a tendency for users to expect quarts from pint pots, and at times I suspect that sufficient margins of output are not being allowed in order to cater for normal requirements, with the result that boilers are operating on peak loads for prolonged periods. Surprisingly enough, my company, which is one of the leading inspecting authorities, has only experienced about three furnace failures (not associated with shortness of water) within the last seven years. These furnace failures took the form of bulging at the lower quarters and about one furnace diameter downstream of the burners. One failure was attributed to a layer of hard silica scale, approximately 0.010 in thick at the water side, whilst in the other failures there was a suspicion of either flame impingement or steam blanketing. Although the author’s experiments show reasonable back-end gas temperatures, experience indicates that much higher temperatures occur in actual practice, and consequently tube plates and the protruding tube ends at the entrance to the first hot pass are vulnerable to overheating and structural damage, particularly on dry-back package boilers, due to the increased heat radiated from refractory linings. I agree with the author’s conclusion that there is a need for more experimental work into the effects of furnace diameter on heat flux, and I also consider that in dry-back package boilers there is a great need to examine back-end conditions, with a view to improving combustion-chamber design and ensuring better methods of tube/tube plate @

D263

attachments which will result in better heat transfer. Experience has shown that where tubes are properly expanded to ensure good contact in the tube holes, and the tubes are welded in with a J type preparation and finished flush, damage from overheating is reduced to a minimum; at least one manufacturer has already adopted this practice. W. Hollick Fellow and K. W. Richardson London The author has attempted to introduce some order into an aspect of design where there has previously been no adequate guidance. The thought that modern designs are satisfactory only if the water-side surface is clean is an alarming one when consideration is given to the difficulties of securing a continuously high standard of water treatment in small industrial boiler plants. The absence of guidance in standards has been evident for many years and some years ago we began to take a serious interest in furnace ratings, the object being to maintain a continuous check on all boilers purchased and, in the absence of definitive knowledge of how far it was practical to increase ratings, to avoid buying boilers which appeared to be particularly ‘advanced’. It was recognized that higher ratings enabled production of cheaper boilers and this showed very clearly in an analysis made some 18 months ago of the capital cost per 1 Ib of steam output from & at 212°F over the preceding 12 years. The analysis showed that the cost had not risen! There is therefore a positive advantage to the purchaser in increasing ratings provided this is not carried too far. The question is ‘What is too far ?’ and this paper goes some way towards answering this question. Ten or more years ago we began to compare ratings, using as a basis the German standard illustrated in Fig. 6. In those days it was common to find a very wide scatter in the figures used, ranging for instance from 0.6 M to 8.5 MBtu/ft h. It is also interesting to note that some boiler designers have roughly doubled the ratings used during that period. Of course, those whose figures were high in the first place have had little scope for ‘improvement’ as is clearly demonstrated in the paper. Early attempts to understand this subject more fully did, we believe, help to dissuade those people who wished to revise B.S. 1971 (18)so as to permit tube metal thicknesses much in excess of 0.875 in. In light of our present knowledge such a move would probably have had disastrous results. Our next big step in knowledge came with an attempt to fire natural gas on some boilers with highly rated furnaces. Furnace tubes did not bulge but there was a great deal of trouble at the rear tube plate ends of the second-pass tubes and the furnace tubes due to the lower radiation transfer with gas flames and hence higher gas temperatures at the furnace-tube exit. Some trouble had already been experienced with tubeend cracking on a few highly rated boilers with oil firing. This suggests that some boiler makers were designing to the limit in this respect and that the increased temperatures with gas firing caused the limit to be exceeded. The convection heat transfer in a furnace tube is small compared with the radiant transfer and the radiant transfer is proportional to the fourth power of the tem-

IMechE 1973

Proc lnstn Mech Engrs Vol187 64/73

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

DISCUSSION ON D. C. GUNN

D264

perature and the surface area of the furnace. It is therefore reasonable to say that the heat transfer in the furnace and hence the temperature of the gases leaving the furnace will have the following approximate relationship :

H = kATG4 or

(

TG = k -

where ,TG is the temperature of gases at furnace exit (“F), A the gas-side surface of water-cooled part of furnace tube (ftz), and H the net heat input to furnace (Btu/h). As H and A are easily determined this gives a simple method of calculating the gas exit temperature if suitable values of k can be determined. It was suggested by the author at a Steam Plant Group informal discussion that the values of k for a rotary burner should be 90 for oil and 120 for gas. Fig. 11 shows experimental results obtained from tests on five different sizes of dry-back boilers (all made by the same boilermaker and fitted with medium-air-pressure burners) with both oil and gas firing. These results indicate that the values of k should be 120 for oil and 150 for gas. While these tests were all carried out on dry-back boilers we cannot see any reason why the furnace-exit temperature should be different on wet-back boilers although the tube-piate and tube-end temperatures are lower because of the cooling effect of the wet-back chamber. However, the T.N.O. report (19)indicates that the heat transfer in a boiler furnace is considerably affected by the burner design. It would seem, therefore, that this method of calculating the exit gas temperature requires different values of k for each different burner. We have adopted the ‘unscientific’method of comparing ratings and avoiding the highest. With increasing years of

operating experience and the additional knowledge of metal temperatures which became available it was possible to set down what were regarded as reasonable figures. Those in current use are given below and it should be emphasized that most of the concern’s plants have reasonably good water treatment and we should certainly suggest lower figures if the water treatment were seriously suspect. The figures used are: Maximum net heat input per foot of diameter: 6.5 to 7.5 MBtu. Preferred ratio of furnace length to diameter: in the range 3-45. (Less than 3 not regarded as good: more than 4.5 not detrimental but not necessary.) Net heat input per cubic foot of furnace volume: 220 000 Btulh. Net heat input per square foot of furnace surface: for oil firing 165 000 Btu/h; for gas firing 140 000 Btu/h. Note that the figures per unit surface are for dry-back boilers :for wet-back boilers the same figures are used but one-third of the wet combustion-chamber surface is added to the furnace area. Application of the above criteria to a recent design for oil firing by the author’s company has given the following figures : Net input per foot of diameter Length/diameter ratio Net heat input per cubic foot Net heat input per square foot

7.91 x lo6 Btu/h 3.63 152 115 164 250 (furnace only)

8

Experimental results 0 Gas f i r i n g Q O i l firing

Net heat input to furnace Btu/h per square foot of surface area

Lines drawn from T = k

(2”’

where T = temperature of gas at furnace exit; H = net heat input to furnace, Btu/h; A = surface area of furnace, ft2. Fig. 11. Net heat input per square foot of furnace plotted against temperature of gas at furnace exit for Economic boiler furnaces 0 IMechE 1973

Proc lnstn Mech Engrs Vol 187 64/73

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

HEAT TRANSFER I N THE FURNACES OF SHELL TYPE BOILERS

Net heat input per square foot

136 167 (furnace++ combustion chamber) The quoted gas temperature at furnace exit is 1802°F and if k = 90 the calculated temperature is 1812°F. It is interesting to note that the Federal Construction Council of the U.S.A. use a similar method for limiting furnace heat input and that their limits are much lower than the ones mentioned above. Section 3 refers to a single corrugated furnace of diameter 3-3 ft and length 24.6 ft having a heated surface of 285 ft2. Since the area of a plain cylinder of the dimensions indicated is 255ft2 we should like to know whether allowance has been made for the corrugations. Section4, paragraph 1,refers to T.N.O. requirements for maximum metal temperatures. Here it may be of interest that we have measured rear tube-plate temperatures higher than 750°F with oil firing and higher still with gas. REFERENCES

(IS) BRITISHSTANDARDS INSTITUTION. B.S. 1971 :1969 Corrugated furnace tubes for cylindric& boilers (London). (19) CLAUS, J. T.N.O. report, November 1970.

A. G. Hotsler and D. M. Lucas Solihull, Warwickshire The heat transfer tests referred to in (I) were carried out jointly by Clarke Chapman-John Thompson Ltd and the Midlands Research Station of the British Gas Corporation. Since these tests British Gas have independently carried out many other trials on different sizes and types of shell boiler fired on both oil and gas. The results have generally confirmed the original work and have been reported in a paper presented to the Institution of Gas Engineers (20). Briefly, it has been found that the total heat transfer and the peak heat flux in the fire tube are lower with gas firing than with oil. The result is that the temperature of the combustion products entering the reversal chamber is higher with gas. The consequent increase in the tube-plate and tube-end metal temperature is, however, small provided there is adequate feed-water treatment. Since the original tests we have also been developing a mathematical model with the aim of helping to interpret and extrapolate the results of field trials. The model has been described fully (21) and consists in splitting the fire tube into a number of plug flow zones, within each of which it is assumed that the temperature, emissivity and convection conditions are uniform. The author’s model is thus essentially the first zone of ours. We do not assume, however, that the combustion is completed within this zone. Detailed measurements (22) in a simulated fire-tube rig show that for a typical burner the combustion is only 70-80 per cent completed in the first diameter. We have found also that the convective heat-transfer coefficient is seriously underestimated by the equation on p. 813. Coefficients have been obtained by experiments in a scale model of a boiler fire tube and these are some eight times higher. If the effects of incomplete combustion and increased convection were included in the author’s model, the calculated flame temperature would be lower. The emissivity necessary to ‘fit’ the model to the measured heat fluxes would then be higher and thus closer to the true value. These detailed changes do not, however, change the

D265

heat fluxes shown in Fig. 8 and it is surely the practical use of this figure in estimating peak heat fluxes that is the most important outcome of the author’s analysis. REFERENCES (20)

(21)

(22)

HORSLER, A. G. and LUCAS,D. M. ‘The firing of shell boilers’, The Gas Council, Research Communication GC195, presented to the 38th Autumn Research Meeting of the Institution of Gas Engineers, London 1972. LUCAS, D. M. and LOCKETT, A. A. ‘Mathematicalmodelling of heat flux and temperature distribution in shell boilers’, 4th Symp. Flames Znd, 1972 (Institute of Fuel, London). RHINES, J. M. ‘Aerodynamic and heat transfer characteristics of proprietary burners fired with natural gas in a cylindrical furnace’, 3rd Members’ Conf. int. Flame Research Foundation 1974. Report No. E240 (British Gas Corporation, Solihull, Warwickshire).

F. E. Lawrence Fellow In defence of B.S. 2790 it must be said that it was never the intention to specify mean metal temperatures because it is not possible to generalize; nevertheless what were believed to be reasonable minimum values were incorporated for guidance. Clause 2.2.2 makes this clear in its wording: ‘The mean metal temperature T used to evaluate the allowable stress f, shall be declared by the manufacturer. The following shall be regarded as minimum requirements, etc.’ I n both Holland and West Germany there are Rules limiting heat input to boiler furnaces, the stated purpose being to avoid overheating problems. I n assessing the benefits of such provisions one must of course be careful to isolate cause and effect, otherwise unwarranted assumptions may be made regarding their efficacy. I n the U.K. there is overwhelming evidence to show that in the vast majority of cases of serious overheating of furnaces in shell boilers the cause is gross shortage of water. This is usually brought about by failure to maintain a working water-level, the result of neglect by the operator or malfunction of automatic controls. It follows that in the normal course of events design heat flux is a much less common cause and it should not be limited by mandate to some arbitrary value. Such action could stifle design flexibility unnecessarily, particularly with respect to the disposition of heat-transfer surfaces and development of firing equipment; for example, it was shown by experiment many years ago that the removal of some smoke tubes from over the crown of the furnace of a particular type of shell boiler reduced the metal temperature significantly under constant conditions of steam output, doubtless because of improved circulation in the area concerned; thus the solution was not to reduce the rating of the boiler, but to first re-examine its design. While manufacturers must obviously make every endeavour to provide reliable boilers and ancillary equipment, experience indicates that operators can provide the only practical solution to the problem of serious overheating by ensuring adequate maintenance as well as appropriate. feed-water and boiler-water treatment. Operational evidence shows that shortage of water in automatically controlled boilers resulting in damage or explosion can be attributed to the following causes. In this context the attention of the reader is drawn to a Technical Data Note (23). Failure to blow through the external float or the control chambers sufficiently often, or failure to carry out the

@I IMechE 1973

Proc lnstn Mech Engrs Vol187 64/73

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

DISCUSSION ON D. C. GUNN

0266

operation correctly so that a clear waterway is proved. Failure to leave the float, or control-chamber isolating valves in the correct working positions after blowing through. Lack of regular maintenance of controls and alarms. Inadequate standards relating to the supervision and testing of controls; occasionally inadequate standard of controls. (Weekly checks should be made by lowering the water level in the boiler and by pressure tests.) The conclusion reachep‘by the author that there is no risk to the furnace metal if the water side is clean, even with the highest heat fluxes indicated in the paper, is valid and the importance of water treatment is rightly emphasized. It is unfortunately true that many operators, after establishing correct treatment principles, spoil their endeavours by failing to maintain the standards set and allowing irregular or incorrect sampling.

Fluid-bed combustion offers one attractive possibility, because here is a system where the average heat flux is also the maximum, It was this thought that stimulated the work for the National Coal Board at B.C.U.R.A. nearly a decade ago into a new approach for firing shell boilers, and later work by the National Coal Board, stimulated by the present fuel situation, should lead to a new generation of shell-type boilers fired by fluidized beds which should become commercially available in the near future. These firing appliances and boilers should, incidentally, confirm what we in the coal industry have long claimed: that coalfired boilers can be as automatic and small as those oiland gas-fired boilers referred to in the paper’s opening paragraph. REFERENCE

(24) RAWLINGS, C. M., RODERICK, D. J. I. and THURLOW, G. G. ‘Note on heat-transfer rates occurring in the furnace tubes of coal-fired shell boilers’, 3. Inst. F. 1962 35, 56.

REFERENCE

(23) H.M. FACTORYINSPECTORATE ‘Safe operation of auto-

matically controlled steam and hot water boilers’, Technical Data Note No. 25.

G . G . Thurlow Fellow

Reading this paper encouraged me to turn up a paper by Rawlings, Roderick and myself, published in 1962 (24). We concluded then from work at the British Coal Utilization Research Association (B.C.U.R.A.) that, for coal-fired shell boilers, local heat fluxes over 100 000 Btu/ ft2 h could occur over the furnace-tube surface, even though the average heat flux was only about 30 000 Btu/ ft2 h. The paper mentions discussions at the Heat Transfer Panel of the R/16 Shell Boiler Advisory Committee of the B.C.U.R.A., a panel of which the author and I share pleasant memories of a high level of technical discussion, often, as in this case, in advance of the times. We concluded at these discussions that ‘local heat fluxes below about 150 000 Btu/ft2 h were unlikely to be troublesome, qrovided that the plate was clean and that no steam blanketing takes place.’ We pointed out in the paper that the tendency (in 1962) was ‘to go for higher and higher heat-release rates . . . particularly for oil-firing’, and we urged the need for very close control of water treatment to minimize scale formation and for further research into levels of heat flux at which steam blanketing is likely to occur. I think I am justified in saying that further work at B.C.U.R.A. and elsewhere has shown that steam blanketing by itself is not a danger but it is gratifying to read the author, after his careful experimental work on oil- and gasfired boilers, reiterating the same note of caution on the importance of a high standard of water control. Would he be prepared to hazard a guess at the frequency of boiler failures due to neglect of this advice ? Have we reached the limit of heat-realease rates in shell boilers ? Obviously one possibility would be to reduce the high local heat fluxes relative to the mean, so making it possible to raise the average heat flux without a corresponding rise in peak value. It is, however, unlikely that a significant change in the heat-transfer pattern is possible with a turbulent diffusion flame as produced by an oil or gas burner. Work at the International Flame Research Foundation and elsewhere has, I think, shown that what can be achieved in this way is limited. Should we therefore be looking for some new approach ?

E. Walker Member The paper collates much useful data into a simple form which could provide an admirable basis for furnace design. It also indicates a requirement for accurate prediction of peak flux in relation to fixed limits when these are specified. This is particularly important in the context of the trend towards more sophisticated design codes. It is difficult, however, to reconcile the 12 per cent accuracy of the mathematical prediction with the fact that, in a given combination of furnace diameter and heat input, changes in burner quarl angle, refractory length and air distribution can have a drastic effect on the peak flux. My company has carried out tank tests, also on a 3.75-ftdiameter furnace, in which calibration of flux meters was checked by measurement of furnace-exit enthalpy in a refractory-lined section of duct. The results of these tests indicate that Fig. 8 over-estimates the peak flux by about 45 per cent over the full range of heat input for this particular furnace and burner configuration. Further evidence is presented by investigation of a case of furnace bulging, caused by a build-up of water-side scale, which occurred some years ago in a twin-furnace 250001b/h boiler operating at a gauge pressure of 350 lb/in2. The furnaces were of 2.75 ft diameter and 8 in thickness and the scale in the region of bulging, about 4 ft from the burner, was 0.1 in thick and contained 35-40 per cent of calcium silicate. Calculation based on determination of the mean metal temperature necessary for bulge initiation indicated a peak flux of 67 000 Btu/ft2 h compared with a predicted flux of 95 000 Btu/ft2 h from Fig. 8. Thermal gradient stress is not mentioned in the paper but this appears to be important at the higher end of the heat-flux and thickness range. Calculations indicate that in a carbon-steel furnace 3 in thick, without scale, the combination of pressure and thermal stress reaches the shakedown limit on the inside surface at a heat-flux cycling between zero and about 95 000 Btu/ft2 h. After some period of service at flux levels above this value the fire-side surface is likely to exhibit fatigue cracks (which would be self-limiting). The maximum heat flux for shakedown to elastic behaviour varies approximately in inverse proportion to the furnace thickness and, as indicated by Fig. 10, it should be possible to design thin furnaces for high peak flux if the water-side surface is maintained in a scale-free condition. 0 IMechE 1973

Proc lnstn Mech Engrs Vol 187 64/73

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

HEAT TRANSFER I N THE FURNACES OF SHELL TYPE BOILERS

I echo the author’s comments on the importance of correct water treatment. As shown by the example above, if scale is permitted to form, even a cautious limit to the peak flux cannot guarantee immunity from bulging. D. C. Gunn Glasgow (Author) D. Billingham raises the interesting question of whether scale forms because of overheating or itself causes the overheating. As more scale forms from more heat input this causes higher metal temperature and both suggestions are true. He then refers to the possible effects on metal temperature of gas-side deposits. These impede heat flow in their locality, but deposits are never uniform so that the heat flow and metal temperature will also be non-uniform, which could create local stress problems. In the tank tests, referred to in section 5, D. Billingham assumes that 50 per cent of the net heat input is transferred in a furnace 2.75 ft in diameter having a heated surface of 81 ft2. With the very high heat input, 19.76 MBtu/h, this assumption cannot be justified, and if the H/vD2criterion is used with Fig. 8, for H/vD2 = 19.76 M/ ~ 2 . 7 5=~835 000 Btu/ft2 h, the corresponding peak heat flux is just 100 000 Btu/ft2 h which compares with the measured peak value of 107000. The average heat flux will be considerably less than this, as shown in Fig. 4. If a very approximate ratio of 1.4 be used, see (3) and (4), a figure of about 70 000 Btu/ft2 h is reasonable. On 81 ft2 this gives a heat transfer of 5.7 MBtu/h or 29 per cent of the gross heat input, not 50 or even 47 per cent as D. Billingham has suggested. The figures from these tank tests are therefore not out of line with the other figures quoted in the paper. The purpose of the paper has been to provide a means of estimating peak heat transfer and hence maximum metal temperature. D. Billingham doubts whether 122000Btu/ft2h is realistic, but theory (Fig. 8) and measurement (Fig. 7) both show that it is. Moreover, G. G. Thurlow later in the discussion mentions a figure of 150 000 Btu/ft2 has having been thought of many years ago! K. S . Chatland, in agreeing that there is a need for correct water treatment, draws attention to circumstances which can arise in practice and prevent the maintenance of the correct boiler-water condition. In the larger plants, fortunate in possessing a laboratory capable of monitoring boiler water, a change of water conditions is noticed and corrected long before any damage is done, but such plants constitute only 10 per cent of the total. Changes in water can and do occur and damage is done. K. S . Chatland remarks that fusible plugs are not now generally used, but I know of at least one case where a fusible plug prevented a nasty accident. The alternative suggested is to measure continuously the temperatures of vulnerable areas of the boiler. Technically this can be done and automatic lockout can be made to occur should the boiler overheat or a thermocouple break. The problem is that such a device is expensive and its additional cost would put the boilermakers standardizing its use in an unfavourable competitive position. A solution would be to require the fitting of the device on all boilers, but an exhaustive series of tests leading to approval of selected types would be necessary. The views of the Inspectorate on the need for such apparatus would be most helpful. I agree entirely with H. P. Heward‘s contention that @

D267

special care is needed in the. selection and adjustment of burners. One has only to study the T.N.O. work (3) and the results appertaining to burner B (Fig. 4) to realize that the burner problem is critical. Fig. 4 shows that burner B (rejected by T.N.O.), whilst giving a high heat flux at the boiler front, which is undesirable, also gives a low heat flux at the outlet end. The T.N.O. work confirms that this lower exit flux corresponds to a lower exit-gas temperature, which should improve the reliability of tube plates. H. I?. Heward draws attention to the relatively few furnace failures that have occurred from causes other than low water, and says that one of these could have been due to steam blanketing. F. E. Lawrence and G. G. Thurlow also raised this point, and in my reply to G. G. Thurlow I suggest that foaming conditions may be important. Flame impingement, with oil firi.- 7, generally results in carbon deposition which impedes heat transfer, but much more needs to be known of the details of flame impingement. I fully agree with H. I?. Heward’s point about flush tubes and prepared welds, but another aspect of tube failure has recently been discovered by British Gas. This is that tube-plate and tube-end temperature is related to the difference in carbon monoxide content between tube inlet and tube outlet. This finding would seem to relate to a process of afterburning brought about by burner imperfections and is obviously of great importance. We are thus brought back to the burner. Its characteristics should be that it does not give a dangerous heat flux at the boiler front and yet allows complete combustion to take place before theend of the furnace. I n other words, it must not shift the hazard from the furnace to the tube plate. This is a subject which is in its early days of exploration, but much of value seems likely to be contained in it. The later points made in the contribution of W. Hollick and K. W. Richardson are dealt with first. They asked whether the heated surface of the Dutch Cornish boiler allows for the increased area due to corrugation. Re-examination of (3) shows that it does. They state that tube-plate metal temperatures exceeding 750°F have been measured. There could be three reasons for this. There could be a genuinely high heat flux, particularly with a dry-back boiler. The very high temperatures recorded in their graph support this suggestion. There could be water-side scale. The rear-tube plate is an area where it may form and be very difficult to remove. It could be an effect of after-burning (see my reply to H. P. Heward and his reference to the work of British Gas). I suspect that in reality the cause of these high tubeplate temperatures is a combination of all three. W. Hollick and K. W. Richardson relate furnace-exit temperature to heat input by TG = / z ( H / A ) ~ ’ ~ and show values for K as 120 for oil and 150 for gas, compared with my suggestions of 90 for oil and 120 for gas. There is a considerable difference which needs exploring. I n the first instance I use the gross and not the net calorific value. This fact goes a small but significant way towards reconciliation. Proc lnstn Mech Engrs Vol187 64/73

IMechE 1973

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

D268

DISCUSSION ON D. C. GUNN

Secondly, I have used tube-inlet temperature, not furnace-exit temperature, and, particularly with wet-back boilers, there is a difference of several hundred degrees. I did not use furnace-exit temperature owing to the need for extensive traversing with a suction pyrometer, whereas at the tube inlet fine-wire thermocouples can be supported inside the tubes and led out at the front of the boiler. A suction pyrometer is, however, the best instrument to use, even at the tube inlet, but it is not often available for use by service and commissioning engineers in the field. If my work had been carried out by suction pyrometer, my constants would have been higher; if the contributors had used the gross calorific value and made their measurements at the tube inlet, their constants would have been lower. Thus there is reason to believe that there is little essential difference between the two independent sets of data. There must, however, be a word of warning about the equation for gas temperature. It is obviously a relation based solely on radiant heat transfer and ignores convection. Whilst this is acceptable for oil where convection is relatively unimportant, for gas it is of greater significance and the equation would be improved by the inclusion of a convection component. I fully agree that burner design and setting is allimportant not only with respect to furnace-exit temperatures but also with respect to tube-inlet temperature. Again, my reply to H. P. Heward is pertinent to burner conditions. The disclosure by the contributors’ furnace-exit temperatures is a valuable record of an important aspect of boiler performance, but it is important to know the number and location of measuring stations and the kind of temperature detector used. British Gas seems to be carrying out more research on shell boilers than any other organization since the days of B.C.U.R.A. and H.M. Fuel Research Station, and the contribution of A. G. Horsler and D. M. Lucas is especially welcome. I agree that their mathematical model is more refined than mine and that one would not, except with a very fierce burner, expect combustion to be complete in one furnace diameter downstream from the burner. The single-zone treatment was, however, within my ability and resources and, perhaps surprisingly, gives a pattern of results which fits the facts. Their estimate of 70-80 per cent complete combustion in the first zone appears to be more realistic. However, any mathematical model must make assumptions, assumptions which can be grossly confounded by the whims and fancies of the engineer in charge of setting up the burner, let alone the imperfections of the burner itself. A. G. Horsler and D. M. Lucas suggest that I have seriously underestimated the contribution of convection, which raises points of interest. I have used the only type of relation known to me for calculating the convective heat-transfer coefficient in a tube (9). Admittedly, the case is concerned with hot gases flowing in a tube, not combustion going on as well, and no account has been taken for recirculation and flow reversal. I look forward with interest to studying (22). A consequence of only 70-80 per cent completeness of combustion in the first zone would be lowering of the value of H/nD2, and hence, from Fig. 8, lowering of the

peak heat flux. Yet there is reasonable correspondence between my measured and calculated heat fluxes as shown in Table 2. Could it be that the true picture is 70-80 per cent completeness of combustion in zone 1, compensated for by convection increased by a factor of 8, but with unaltered radiating properties ? Recalculation of equation (1) on p. 813 with the contributors’ enhanced convection component will prove instructive. I n reply to F. E. Lawrence I would say that the paper was intended to provide guidance on the estimation of maximum metal temperatures in the furnaces of boilers. T o this subject the U.K. has, until now, made little contribution, it being left to Holland and West Germany to incorporate in their standards the necessary information. In the absence of such guidance, people will tend to use the minimum requirements specified, which, as Fig. 10 indicates, are generally inadequate. It is a weakness of British and other Standards that the derivations of the various data and formulae have been lost to posterity, a fact which makes difficulties for those responsible for the reconciliation and revision of standards. Thus the reasons why certain minimum metal temperatures have been laid down in B.S. 2790 are not now known, and may well relate to boilers of a past age. The U.K. is at present providing the secretariat for the preparation of an 13.0. standard on shell boilers. The proposal is largely based on B.S. 2790, but a means of calculating maximum metal temperatures, based on my paper, is being offered as an alternative to the Dutch and German approaches. It is interesting to note that only the Dutch, so far, have given any information concerning the derivation of their data. The Germans have made a rule, but given no reasons for it. F. E. Lawrence, as a prominent member of the Inspectorate, is in an authoritative position to ascertain the causes of furnace failure and I do not disagree that the majority are caused by low water. There have been failures, however, where the cause has been less clear and there is some evidence to show that thermal impedance due to scale (sometimes very thin silicate scale, or even foam) in the locality of peak heat fluxes has certainly contributed sometimes to bulging and sometimes to weld failure at the junction of the furnace and tube plate. A standard should show how a safe reliable boiler can be made. Surely metal temperatures and therefore heat flux are of considerable importance in this regime ? I agree, however, that excessive heat flux in furnaces is a less common problem than leakage at the hot tube plate, and that an attempt to safeguard the furnace might shift trouble further down the boiler. This is another subject, but has been raised elsewhere in the discussion, particularly in my reply to H. E. Heward. I cannot but agree with F. E. Lawrence’s point that, overriding the best technical skills, :A the ever-present danger of human error. I am grateful t o G. G. Thurlow for reminding me of the happy and constructive proceedings of the Heat Transfer Panel. It is certainly true that much of the work of that body has led up to the present paper. The figure quoted by him, 150 000 Btu/ft2 h, as being a ‘safe’ limit to heat flux in a clean boiler lies close to the predicted maximum heat flux in a shell boiler with present methods of firing oil (see curve 5 in Fig. 8). On the subject of steam blanketing I recollect that an

Proc lnsln Mech Engrs Vol 187 64/73

@ IMechE 1973

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016

HEAT TRANSFER I N THE FURNACES OF SHELL TYPE BOILERS

attempt was made at B.C.U.R.A. to demonstrate this phenomenon by heating an open copper pot, containing water, with a group of oxy-acetylene burners, At a heat flux of 400 000 Btu/ft2 h, film boiling had not taken place. On the other hand I was later concerned with furnace-intank experiments, the water for which was drawn from a river. Small fish penetrated the screen at the pump inlet, were cut in pieces by the impeller, and cooked in the tank. As time went on the tank water became very foamy (and smelly) and some extraordinarily high metal temperatures were measured. Unfortunately the tank was cleaned out and the planned experiment resumed, otherwise some useful information on heat transfer to a foamy liquid might have been obtained. It could well be that under foaming conditions, by no means uncommon in boilers, a condition of thermal impedance can exist which is akin to film boiling. A few furnaces in boilers, known to be full of water at the time, have bulged in the bottom half, in localities where foam might have been trapped. F. E. Lawrence draws attention to a case which might be very similar, but the top half of the furnace was affected. It should be remembered that in a boiling, foamy, liquid the foam extends below the surface, as well as above it. See also (25). Concerning the number of boiler failures due to lack of water quality control I must decline to ‘hazard a guess’. The Inspectorate will have access to authoritative records of failure. It was, however, the need to emphasize the importance of water treatment that prompted the Association of Shell Boilermakers to produce (16). G. G. Thurlow has mentioned fluid-bed combustion. It is difficult to visualize such a system in a horizontal shell boiler, and a vertical shell boiler has its output limited by the small area of water surface available for steam disengagement. Just as the present-day shell boiler bears little resemblance to the Lancashire boilers of former years, so the fluid-bed boiler of the future is likely to differ entirely in geometry from present-day boilers. With such a difference in combustion system, it would seem unwise to put ‘new wine into old bottles’ and far better to make the bottle suit the wine. The real points, however, are to identify the advantages (which must in the end be mainly economic) of fluid-bed combustion and to encourage the industry to change its ideas. I agree fully with E. Walker about the importance of burner details in determining peak heat flux. An object of the paper was to determine what these details are likely to be under the worst conditions, and many of the burner faults today, e.g. asymmetric air distribution, tend to transform possible trouble with high heat flux in the furnace into actual trouble further down the boiler. In the sample of measurements I used (the largest I could obtain) the calculated and measured results are in fact within 12 per cent as shown in Table 2. With reference to Fig. 7, however, there is one outstanding point, that for burner B at HID = 2 MBtu/h, where the error is 18 per cent. If reference (3) is examined it will be seen that burner B showed undesirable characteristics which were overcome by altering the quarl shape. This emphasizes E. Walker’s points about burner details. He suggests that the paper over-estimates the heat flux by 45 per cent. How can this be when measurements (not only mine) confirm the calculated results much more closely than this 3

D269

There could have been a difference in the measuring techniques used. I used differential thermocouples buried in the furnace plate to accurately known depths; where these can be used I know of no better method. What sort of flux meter did E. Walker use ? The estimation of furnaceexit enthalpy could be very difficult when one considers the gas stratification that exists in this locality. Not only is it necessary to evaluate temperature, but gas composition and quantity are also needed at every temperature station, and none of these is feasible without instrumentation approaching the scale of the laboratories of the International Flame Research Foundation. I therefore leave the question open: ‘Whose measurements are nearer the mark, E. Walker’s or mine ?’ Another possibility is that E. Walker used burners in entirely different trim from mine. I adjusted the burners to give the fiercest conditions. He refers to a bulged furnace in which, under the scale conditions existing, a heat flux of only 67 000 Btu/ft2 h would have been sufficient to cause the bulge. Surely any higher heat flux would also have caused the bulge, and 67 000 Btu/ft2 h is the average heat flux in furnaces, not the peak. The paper was specialized and concerned with stresses. Stresses are all-important; now that we have a means of predicting temperatures, surely the next and essential step is to discuss the stresses arising from them. Perhaps E. Walker could oblige the Institution in the not too distant future ? In conclusion I wish to thank all contributors for their careful study of the paper; I have much enjoyed reading their views, whether or not they coincide with my own. The following points stand out.

D. Billingham, an authority on water treatment, has enlarged on the need for even better methods of water treatment than those at present contemplated. K. S. Chatland has drawn attention to the possibility of using a temperature-sensitive instrument to safeguard the boiler should overheating of vulnerable parts occur. H. E. Heward has drawn attention to problems arising with tube plates and the need for special arrangements in this part of the boiler, pointing out, in common with F. E. Lawrence, that furnace failures due to causes other than low water have been few in number. W. Hollick and K. W. Richardson have presented factual data from the field for combustion-chamber gas temperatures. There is, according to A. G. Horsler and D. M. Lucas, the probability of a serious underestimate of convective heat transfer. They also draw attention to a more sophisticated mathematical model shortly to be presented to the International Flame Research Foundation. G. G. Thurlow has raised the fascinating issue of fluidized beds in boilers. E. Walker has drawn attention to the importance of burner details and furnace stresses. Surely here are the bases of a variety of useful and topical papers for the Steam Plant Group ? REFERENCE

(25) HILLIER, H. ‘Feed distribution and hunting in marine water-tube boilers’, Proc. Instrz mech. Engrs 1947 156, 139-75.

@ IMechE 1973

Proc lnstn Mech Engrs Vol 187 64/73

Downloaded from pme.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016