fp7-sme-2010-1 262205/ intheat page 1 of...

FP7-SME-2010-1 262205/ INTHEAT

29/11/2011

of 30

Project no.: 262205

Project full title: Intensified Heat Transfer Technologies for Enhanced Heat Recovery

Project Acronym: INTHEAT

Deliverable no.: D1.2

Title of the deliverable: Report on experimental fouling investigation and CFD research on

heat transfer enhancement

Contractual Date of Delivery to the CEC: 30/11/2011

Actual Date of Delivery to the CEC: 30/11/2011

Organisation name of lead contractor for this deliverable: UNIBATH

Author(s): Barry Crittenden, Mengyan Yang

Participants(s): P1, P2, P3, P6, P7, P8, P9

Work package contributing to the deliverable: WP1

Nature: R

Version: 1.0

Total number of pages: 30

Start date of project: 1st December 2010

Duration: 24 months

Project co-funded by the European Commission within the Seventh Framework Programme (2007-2013)

Dissemination Level

PU Public

PP Restricted to other programme participants (including the Commission Services)

RE Restricted to a group specified by the consortium (including the Commission Services)

CO Confidential, only for members of the consortium (including the Commission Services) X

Abstract:

Fouling and scaling experiments have been conducted on crude oils and water, respectively, using

the batch stirred cell at the University of Bath. The effects of temperature and velocity on the

fouling and scaling phenomena have been interpreted with the help of CFD modelling using

COMSOL. Crude oil fouling threshold conditions have been obtained. They may provide useful

guidance to the avoidance of fouling by optimising operational parameters. CFD simulations of

fluid flow and heat transfer in tubes fitted with inserts and flows around a wired test surface have

helped to gain understanding of the effects of the geometric alternatives on turbulence

enhancement, and hence on fouling mitigation techniques.

of 30

FP7-SME-2010-1 262205/ INTHEAT Confidential

THEORETICAL AND EXPERIMENTAL INVESTIGATION OF FOULING

Fouling; heat transfer, crude oil, scaling, heat exchangers, threshold plot, mathematical models.

Table of Contents

BIBLIOGRAPHY ............................................................................................................................... 3

NOMENCLATURE ............................................................................................................................ 4

1. EXECUTIVE SUMMARY ............................................................................................................. 5

2. INTRODUCTION ........................................................................................................................... 6

3. EXPERIMENTAL INVESTIGATION ........................................................................................... 7

3.1 APPARATUS, MATERIALS AND EXPERIMENTAL METHOD ......................................... 7

3.2 RESULTS AND DISCUSSION - CRUDE OIL FOULING .................................................... 9

3.2.1 Effect of temperature and velocity on fouling rate –fouling threshold ............................................. 9

3.2.2 Fouling threshold conditions .......................................................................................................... 11

3.2.3 Effect of turbulence enhancement on fouling ................................................................................... 12

3.3 RESULTS AND DISCUSSION – CASO4 CRYSTALLISATION FOULING ....................... 12

3.3.1 Mitigation effect of wire attachment on fouling .............................................................................. 13

3.3.2 Influence of surface materials ......................................................................................................... 14

3.3.3 Asymptotic fouling ........................................................................................................................... 14

4. CFD SIMULATION AND MODELLING ................................................................................... 14

4.1 CFD SIMULATION OF FLOW IN THE STIRRED CELL WITH A WIRED PROBE .......... 15

4.1.1 Effect of shear stress ....................................................................................................................... 15

4.1.2 Effect of temperature ...................................................................................................................... 17

4.2 CFD SIMULATION FOR HEAT TRANSFER IN TUBE FITTED WITH AN INSERT ....... 18

4.2.1 Temperature field in tube fitted with insert ..................................................................................... 18

4.2.2 Average heat transfer coefficient .................................................................................................... 20

4.3 RESEARCH ON FOULING MODELS .................................................................................. 22

4.3.1 Fouling rate and fouling thresholds ................................................................................................ 22

4.3.2 Compensation plot ........................................................................................................................... 24

4.3.3 Modelling of fouling induction periods ........................................................................................... 24

5. CONCLUSION .............................................................................................................................. 26

6. REFERENCES............................................................................................................................... 27

of 30


BIBLIOGRAPHY

The following bibliography has been assembled during the project to provide a wide-ranging

account of the principles underlying the problem of fouling as well as the approaches which have

been adopted for its mitigation. The research element required in finding solutions to the problem is

significant, and accordingly the bibliography is invaluable in assisting organisations to find

practical solutions to individual fouling problems.

Books

• Somerscales, E. F. C. and Knudsen, J. G., (1981), Fouling of Heat Transfer Equipment,

Hemisphere Publishing Corporation, Washington.

• Suitor, J. W. and Pritchard, A. M., (1984), Fouling in Heat Exchange Equipment, American

Society of Mechanical Engineers, New York.

• Garrett-Price, B. A., Smith, S. A., Watts, R. L., Knudsen, J. G., Marner, W. J. and Suitor, J.

W., (1985), Fouling of Heat Exchangers: Characteristics, Costs, Prevention, Control and

Removal, Noyes Publications, New Jersey.

• Melo, L. F., Bott, T. R. and Bernardo, C. A., (1987), Fouling Science and Technology,

Kluwer Academic Publishers, Dordrecht.

• Bott, T. R., (1990), Fouling Notebook, Institution of Chemical Engineers, Rugby.

• Bott, T. R, (1995), Fouling of Heat Exchangers, Elsevier Science & Technology Books.

• Müller-Steinhagen, H., (2000), Heat Exchanger Fouling: Mitigation and Cleaning

Technologies, Institution of Chemical Engineers, Rugby.

Major Biennial Conferences (available on-line)

• Heat Exchanger Fouling and Cleaning: http://www.heatexchanger-fouling.com/index.htm

Commercial Reports

• IHS ESDU, Heat Exchanger Fouling in the Pre-Heat Train of a Crude Distillation Unit,

Report ESDU 0016, London (ISBN: 978 1 86246 119 2).

• IHS ESDU, Fouling in Cooling Systems Using Seawater, Report ESDU 03004, London

(ISBN: 978 1 86246 220 5).

• IHS ESDU, Fouling in Cooling Systems Using Fresh Water, Report ESDU 08002, London

(ISBN: 978 1 86246 618 0).

of 30


NOMENCLATURE

A Pre-exponential factor in Arrhenius expression m2K kJ

-1

A Dimensioned constant in fouling model

B Dimensioned constant in fouling model

C Dimensioned constant in fouling model

Cb Foulant precursor concentration in the bulk kg m-3

Cf Drag coefficient

E Activation energy or apparent activation energy kJ mol-1

Cε1 Parameter in k-ε turbulence model

Cε2 Parameter in k-ε turbulence model

f Friction factor

k1 Rate constant in induction period model s-1

k2 Removal rate constant in induction period s-1

keff Effective thermal conductivity W/(K m)

ko molecular thermal conductivity W/(K m)

kT turbulent thermal conductivity W/(K m)

u Velocity ms-1

u* Friction velocity ms-1

R Universal gas constant kJ mol-1

K-1

Re Reynolds number

Rf Fouling resistance m2K kW

-1

Rf*

Asymptotic fouling resistance m2K kW

-1

Tf Film temperature K

Ts Surface temperature K

t Time s

Greek symbols

α Dimensioned constant in fouling model

β Constant in fouling model

β Time constant s-1

γ Dimensioned constant in fouling model

θ Fractional surface coverage

µ Dynamic viscosity Nm s-2

ν Kinematic viscosity m2 s

-1

ρ Fluid density kg m-3

τ Shear stress N m2

κ Turbulent energy m2/s

2

� Turbulent kinetic energy dissipation rate m2/s

3

η Dynamic viscosity Pa.s

ηT Turbulent dynamic viscosity Pa.s

σε Parameter in k-ε turbulence model

σk Parameter in k-ε turbulence model

of 30


1. EXECUTIVE SUMMARY

Fouling experiments have been carried out over a wide range of experimental conditions using a

batch stirred cell system which is flexible and easy to operate up to pressures of 30 bar and surface

temperatures of 400oC. The system is suitable for use with hydrocarbons such as crude oils and with

aqueous systems when CaSO4 scaling can be studied. The outside surface of the heated test probe

can be modified to study heat transfer and fouling on certain configurations of enhanced surface.

Linear and falling fouling rates are observed, dependent on the system under study. Induction

periods can also be studied. Once fouling has progressed to a significant level, the cell can be used

to study negative fouling rates by changing the processing conditions in a carefully controlled

manner. In this way, the thresholds of surface temperature and surface shear stress can be found,

below which fouling becomes insignificant. Flow and heat transfer in the batch stirred cell are

complex, particularly when enhanced surfaces are used. Computational fluid dynamics (CFD) has

therefore been used to model the fluid flow and heat transfer, and hence to obtain local distributions

of surface temperature, heat transfer and shear stress. CFD has also been used to study theoretically

the fluid flow and heat transfer that arise when wire matrix enhancement devices such as hiTRAN®

inserts are used in plain round tubes.

A new concept developed within this project has been that of the equivalent velocity/Reynolds

number. It allows a fouling model developed for bare round tubes to be extended for use with more

complex geometries. For example, it has been demonstrated that the fouling models of Ebert and

Panchal (1997) and Yeap (Yeap et al. 2004; Yang and Crittenden 2011) can be adapted successfully

to correlate the fouling data of a crude oil (Crude A) obtained using the batch stirred cell, as well as

for tubes fitted with hiTRAN®

inserts. Recently within this project, the equivalent

velocity/Reynolds number approach has been adopted for modelling fouling in plate heat

exchangers by SODRU. Fouling threshold conditions can now therefore be predicted successfully,

auguring well for the development of successful strategies to mitigate the highly energy consuming

fouling problem in systems where intensified heat transfer techniques are being adopted.

CFD simulation confirms that the average heat transfer coefficient for a tube fitted with an insert is

much higher than that for the bare tube operated under the same conditions of surface wall

temperature and average velocity. The increase in the heat transfer coefficient when an insert is

used means that the temperature in the shell side of an exchanger can be reduced for a given thermal

duty, so helping to reduce the fouling problem. The distribution of local shear stresses which have

an important impact on fouling rates can also be predicted by CFD simulation. The CFD method for

predicting the average heat transfer coefficient is valuable in providing critical information on the

design of a heat exchanger which comprises tubes fitted with inserts. The CFD and heat transfer

simulation provides a valuable tool in studies of the effect on fouling of the local fluid temperature

near the wall. This is especially the case for tubes fitted with inserts or others with other types of

irregular geometry.

The generic model which has been developed for fouling induction periods now makes it possible

to describe the fouling process from the start of the induction period up to the steady fouling rate

stage using a single and simple mathematical expression. The proposed term t0.5 which is the time to

reach 50% of the maximum surface coverage, θmax, provides a practical measure of the length of the

induction period. Tested on experimental data for crude oil fouling, calcium sulphate fouling and

whey protein fouling, the model quantitatively describes the influence of the surface temperature on

the length of the induction period for various systems. The model also describes in a semi-

of 30


quantitative manner the influence of velocity on the induction time.

2. INTRODUCTION

Fouling concerns the formation of unwanted material on heat transfer surfaces. In the full range of

processing industries, fouling creates a chronic operational problem that compromises energy

recovery and environmental welfare. Problems manifest themselves as loss or reduction in

production, increased energy consumption, increased pressure losses, flow maldistributions, anti-

fouling chemical costs, cleaning costs, and so on (Bott, 1990). Mitigation of the problem is

frequently restricted by inadequate detailed knowledge of the underlying mechanisms (Ishiyama et

al., 2009), even though the basic principles have been understood for some time (Melo et al., 1987).

Fouling can be classified into scaling (crystallization), reaction, corrosion, freezing, sedimentation,

biological, and so on. In many industrial situations, more than one type may occur simultaneously

depending on the fluid being processed and the operating conditions, thereby making each fouling

situation almost certainly unique. This makes the identification of fouling mitigation strategies a

particularly intractable problem. Indeed, it makes modelling for design and operational strategies

even more challenging.

Ever increasing energy costs have led to the pursuit of heat integration approaches in the process

industries in order to recover as much heat as possible from the product streams so as to improve

energy efficiency. Commonly, multi-pass shell and tube heat exchangers are used as heat recovery

units, particularly in oil refineries. Due to the complex nature of crude oils, heat exchangers in oil

refinery crude preheat trains are prone to fouling but the fouling process is slow with a large time

constant. Crude oil fouling is generally believed to be caused by impurities in the crude oil such as

corrosion products, water and salt, the precipitation of insoluble asphaltenes, as well as the thermal

decomposition, or auto-oxidation, of reactive constituents in the oil.

Research using actual plant data is slow, subject to a variety of logistical and operational

requirements which do not lend themselves well to fundamental scientific studies (Crittenden et al.,

1992), and can create difficulties in the interpretation of the thermal data (Takemoto et al., 1999). A

number of laboratory methods have been developed to study liquid phase fouling (Epstein, 1981)

including for crude oils the use of the stirred batch cell (Eaton and Lux, 1984), or the use of a

recycle flow loop with either a tubular cross section (eg Crittenden et al., 2009) or an annular cross

section (eg Watkinson and Wilson, 1997; Bennett et al., 2009). For the current project, fouling

experiments at the University of Bath have been carried out using a batch stirred cell. The cell

design follows closely that of Eaton and Lux (1983, 1984) and was chosen since it offers

extraordinary flexibility. Crude oils or water based fluids can be changed easily, as can the fouling

fluid chemistry (eg by adding asphaltenes, metal salts, etc). Computational fluid dynamics (CFD)

software also allows the thermal and fluid flow characteristics of the complex batch stirred cell

geometry to be predicted and validated (Yang et al., 2009a).

In-tube inserts, such as hiTRAN®, have been shown to be effective in mitigating crude oil fouling

and enhancing heat transfer (Crittenden et al., 1993; Ritchie and Droegemueller, 2008). Increasing

interest in their use in such applications is being shown by the oil industry (Krueger and Pouponnot,

2009) as well as by the water industries (Bott, 2001; Wills et al., 2000). A good review of the

applications and benefits of tube inserts in heat exchangers is provided by Ritchie and

Droegemueller (2008). Nonetheless, the use of inserts creates a challenge in the design of a heat

exchanger, due to the insufficient understanding of the fouling behaviour and the lack of practical

of 30


methods for the estimation of some critical design parameters, the heat transfer coefficient in

particular. Fouling of the heat exchanger depends, amongst other things, on two key operational

parameters, namely the wall shear stress and the surface temperature. The use of inserts, moreover,

raises a challenge in the application of fouling models, namely in the determination of the Reynolds

number and the wall shear stress. Indeed, current fouling models are not capable of taking into

account the complex variation of the surface shear stress along the length of the insert. Whilst the

wall shear stress is easily calculated for bare round tubes using the friction factor approach, this

method cannot be used with hiTRAN® inserts. CFD simulation can offer a possible solution to this

challenge, such that a suitably modified fouling model can then be used to predict the fouling rate

and threshold conditions for tubes with and without inserts fitted. This approach can be further

developed in terms of the equivalent velocity/Re number, which is defined to be the velocity/Re

number in a bare tube that gives the same wall shear stress in a tube of the same internal diameter

fitted with inserts and operating at a different average fluid velocity.

Part of the fouling process is induction, initiation or delay in which no fouling appears to take place.

The subject is poorly understood and poorly modelled. Better knowledge might mean that induction

periods could be extended indefinitely. Accordingly, a generic model of the fouling induction

period has been developed and extended to a variety of fouling systems (Yang et al. 2009). The

model allows the effects of both surface temperature and velocity on the induction time to be

studied and interpreted.

3. EXPERIMENTAL INVESTIGATION

Experimental investigations for both crude oil fouling and calcium sulphate scaling have been

carried out using the batch stirred cell. The effect of surface enhancement has been studied using

wires attached to the otherwise plain heated surface. The effects of the wires on heat transfer and

surface shear stress have been simulated using CFD.

3.1 APPARATUS, MATERIALS AND EXPERIMENTAL METHOD

The general arrangement of the cell is shown in Fig. 1. The cell comprises a pressure vessel made

in-house from a block of 304 stainless steel, together with a top flange. The base of the vessel

houses an upwards pointing test probe heated internally by a cartridge heater, the heat flux from

which is controlled electrically (Fig. 2a). The fluid, crude oil or aqueous solution (≈ 1.0 litre), is

agitated by a downwards facing cylindrical stirrer mounted co-axially with the test probe and driven

by an electric motor via a magnetic drive. External band heaters are incorporated to provide initial

heating to the vessel and its contents. An internal cooling coil uses a non-fouling fluid (Paratherm)

to remove heat at the rate that it is inputted via the cartridge heater during the fouling run. The

vessel is fitted with a pressure relief valve and there is a single thermocouple to measure the fluid

bulk temperature. A wire nest comprising 8 vertical wires of 0.7 mm fabricated by Cal Gavin can be

attached to the probe, allowing the effects of attached wires on turbulence, heat transfer and fouling

to be evaluated. Figure 2b shows this configuration schematically. The actual number of wires used

is not the same as shown in the figure.

The cell can be sparged with various gases, eg oxygen and nitrogen, and the heat transfer surface is

easily inspected and changed. Computational fluid dynamics (CFD) software also allows the

thermal and fluid flow characteristics of the complex batch stirred cell geometry to be predicted and

validated (Yang et al., 2009a). The heated probe surface for the crude oil fouling experiments was

of 30


refinery grade mild steel. In order to evaluate the influence of surface properties on CaSO4 scaling,

three new probes were made using mild steel, copper, and stainless steel.

The properties of the crude oil used in the fouling experiments are listed in Table 1. The aqueous

CaSO4 solution was prepared by weighing 4.10 g Ca(NO3)2 and 8.00 g Na2SO4·10HO2 and

dissolving each salt in 0.5 L of distilled water. Then the two solutions were mixed slowly with

agitation. The solution made in this way contains 3.4 g CaSO4 in 1 L water.

t wb t ws

t bulk

© 2008 University of

Bath, England

t wm

Fil Level

Fig. 1 The batch stirred cell

Fig. 2a Heated test probe Fig. 2b Probe with wire nest

twb, twm, and tws: embedded thermocouples

of 30


Table 1 Properties of the Crude Oil Blend Tested

Crude A

API 27.5

Viscosity (cst) @ 80 oC 15

Viscosity (cst) @ 260 oC 1.74

Total Sulphur (% wt) 2.82

Iron (ppm) 4

Nickel (ppm) 42

Vanadium (ppm) 226

Saturates (%) 28.36

Aromatics (%) 56.87

Resins (%) 6.8

Asphaltenes (%; IP143) 8

CII 0.56

3.2 RESULTS AND DISCUSSION – CRUDE OIL FOULING

In virtually all experiments to date, the fouling resistance has been found to vary linearly with time.

An induction period is usually seen when a well-cleaned probe is used, but not when the surface has

not been cleaned thoroughly. A typical crude oil fouling resistance curve is shown in Fig. 3.

Fig. 3 Typical fouling curve for Crude B

3.2.1 Effect of temperature and velocity on fouling rate – fouling threshold

Keeping the stirrer speed constant, but increasing the initial surface temperature by increasing the

power input to the cartridge heater results, as expected, in a higher fouling rate. Figure 4 shows the

effect of surface temperature on fouling rate for four stirrer speeds.

Previous work (Young et al. 2011) has demonstrated the effect stirred speed, or the wall shear stress

has on the fouling rate. That is, keeping the surface temperature constant but increasing the stirrer

speed results in a lower fouling rate. The effect of surface shear stress on the removal of deposits

has now been studied in a different, and novel, manner. After a significant of fouling has been

accumulated, the stirrer speed is then increased and maintained constant at the elevated value for a

reasonably long period whilst maintaining the other key parameter, namely surface temperature,

constant. Under such circumstances, it becomes possible to observe negative fouling rates, as

-0.02

0

0.02

0.04

0.06

0.08

0.1

0.12

0 1 2 3 4 5 6 7

Time (hour)

Rf

(Km

2/k

W)

Rf = 0.0177 t - 0.0185

of 30


shown for example in Figure 5. The power input to the cartridge heater shown in this figure

confirms that the heating power, and hence heat flux, has been maintained constant.

Fig. 4 Effect of surface temperature on fouling rate

Fig. 5 Negative fouling rate at increased stirrer speed

Figure 6 summarises the new method of obtaining the fouling threshold when the fouling rate is

equal to zero. The locus of surface shear stress and surface temperature at which the fouling rate

becomes zero is obtained easily from Figure 6. The surface shear stress for any given stirrer speed is

obtained by CFD simulation (Yang et al. 2009a). The surface temperature is obtained by

measurement.

Fig. 6 Fouling rate against surface shear stress for various initial surface temperatures

0.00E+00

1.00E-09

2.00E-09

3.00E-09

4.00E-09

5.00E-09

6.00E-09

610 630 650

dR

f/d

t (m

2K

/J)

Ts (K)90 rpm 160 rpm 300 rpm 400 rpm

y = -1.9631E-05x + 4.4701E-05

-0.00018

-0.00014

-0.0001

-0.00006

-0.00002

0.00002

0 2 4 6 8 10

Time (hour)

Rf (m

2K

/W )

100

120

140

160

180

200

220

240

Po

wer

(W)

Fouling Resistance Power

-1.00E-08

-5.00E-09

0.00E+00

5.00E-09

1.00E-08

0 0.5 1 1.5 2

Surface shear stress (Pa)

Fo

uli

ng

ra

te (

m2K

/J)

600 K 610 K 620 K 630 K 640 K 650 K 660 K

of 30


3.2.2 Fouling threshold conditions

The concept of fouling threshold conditions, which was introduced by Ebert and Panchal (1997), is

particularly interesting since if the loci of surface temperature and surface shear stress for which

fouling will not occur can be found, then the fouling problem is, in principle, solved. In practice,

fouling threshold conditions, if identified, may provide a guide to avoid fouling or at least to

minimise the impact of fouling by operating a heat exchanger under the non-fouling conditions.

Normally, the fouling threshold would be determined by extrapolating plots of fouling rate versus

surface temperature (at constant shear stress) back to the point at which no fouling occurs (Panchal

et al., 1997; Knudsen et al., 1997). To obtain reliable threshold data in this way, it is necessary to

carry out a large number of fouling runs, including some tests being run at very low fouling rates to

improve the accuracy of locating the zero fouling conditions. Experiments in which very low

fouling rates need to be studied would, however, be extremely time-consuming. In the present work,

given that negative fouling rates can be observed by judicious choice of operating conditions after a

deposit has been laid down on the surface, it now becomes possible using the new method described

in 3.2.1 to identify the fouling threshold conditions by interpolating the plots of fouling rate –

surface temperature (at constant shear stress) to find the points at which the fouling rate becomes

equal to zero. Figure 6 shows fouling rate plots, both positive and negative, at a series of shear

stress values.

The threshold conditions are easily obtained by interpolating the plots shown in Figure 6 when the

fouling rate equals zero. Figure 7 shows the fouling threshold conditions for the four surface

temperatures for which the fouling rate – surface shear stress curve crosses from being a positive

fouling rate to a negative one: 620K (347°C), 630K (357°C), 640K (367°C), and 650K (377°C).

The shear stress values were obtained by CFD simulation based on the physical properties of the

crude oil, the stirrer speed and the bulk temperature used in the experiments (Yang et al., 2009;

Young et al., 2011). Figure 7 seems to show that the fouling rates, when plotted in this way,

decrease linearly with increasing shear stress. It is also worth noting from Figure 6 that the

gradients of the plots of fouling rate against shear stress seem to be constant regardless of the

surface temperature. As a comparison, results from the new interpolation method have been

compared with those from the more time-consuming extrapolation method (reported previously by

Young et al. (2011)) in which zero fouling rates were obtained by extrapolating fouling rate data

back to zero. The comparison is shown in Figure 7.

Fig. 7 Threshold temperature

◊: this work; ●: previous work (Young et al., 2011)

2 4 0

2 6 0

2 8 0

3 0 0

3 2 0

3 4 0

3 6 0

3 8 0

4 0 0

0 0 .5 1 1.5 2

Shear stress (Pa)

Th

resh

old

tem

pera

ture

(°C

)

of 30


Agreement of the two methods is surprisingly good, bearing in mind the previous method

potentially suffered from inaccuracy as the fouling rate curves were extrapolated back to the origin.

Any differences could be due not only to difficulties in extrapolation but also to the fact that the

crude oil used in this work has been subject to fouling for relatively long periods, and accordingly,

some its properties may have changed slightly.

Although the threshold conditions have been obtained using a batch stirred cell system, it could be

argued that they might not bear close resemblance to the industrial situation. However, to counter

this argument it should be borne in mind that they would have a significant value in understanding

the industrial situation provided that a fouling rate for a particular crude oil was determined solely

by the surface temperature and the surface shear stress. That is, the fouling behaviours should be

similar at the same temperature and under the same shear stress, i.e. at the same equivalent

Reynolds number regardless of the geometries of the surface (Yang et al., 2009a).

3.2.3 Effect of turbulence enhancement on fouling

Fins and/or wire inserts fitted in a flow path may enhance the fluid turbulence, and hence increase

the wall shear stress. This can effectively mitigate fouling. To investigate this effect, a test probe

fitted with the wire nest shown schematically in Figure 2b was used in the crude fouling

experiments. Figure 8 shows a typical fouling curve using this probe.

Fig. 8 Fouling resistance versus time using a wired probe

Test condition: bulk temperature 258°C; surface temperature 399 °C

stirring speed 160 rpm, heat flux 79 kW/m2

The fouling data is seen to be rather scattered. This could be due to irregular removal of the foulant

from the surface under elevated shear stress due to presence of the wires. Table 2 lists the fouling

rates and shows that the fouling rate using a wired probe is lower even at higher temperatures when

compared with that using a bare probe.

3.3 RESULTS AND DISCUSSION – CaSO4 CRYSTALLISATION FOULING

Experiments on calcium sulphate scaling have been included for three reasons. Firstly, the aqueous

solution provides much more certainty in its composition than crude oil. Secondly, experiments are

y = 5.47E-10x - 8.37E-06

-0.000015

-0.00001

-0.000005

0

0.000005

0.00001

0.000015

0 5000 10000 15000

Fo

uli

ng

re

sis

tan

ce

(m

2K

/W)

Time (S)


carried out under much milder conditions than those for crude oil. Thirdly, experiments can be

carried out more quickly, so that an understanding of the effects of process parameters, including

enhancement of the heat transfer surface, can be gained more quickly. Fouling experiments were

conducted using both bare and wired probes made of mild steel, copper and stainless steel.

Table 2 Fouling Rate C

Probe

Bare probe

Wired probe

Wired probe

Bulk temperature 258

3.3.1 Mitigation effect of wire attachment

Figure 9 shows the increase in surface

stirring speeds with the calcium sulphate system

directly proportional to the rate of fouling in the batch stirred cel

3, demonstrate that the fouling rate, which is proportio

probe than for the bare probe. However the mitigating effect of the wires seems

significant as for the case of crude oil fouling.

expected, lower at a high stirring speed.

Fig. 9 Surface temperature increase with time due to CaSO

♦: bare probe

▫: wired probe, 130 rpm

Bulk temperature: 55

Table 3 Summary of CaSO

Fouling rate on bare probe

Fouling rate on wired probe

of 30



heat transfer surface, can be gained more quickly. Fouling experiments were


ouling Rate Comparison for Crude Oil – Bare Probe vs Wired P

Surface temperature

(°C)

383.7

385.5

399.3

Bulk temperature 258 °C; Stirring speed: 160 rpm; Crude A

attachment on fouling

increase in surface temperature with time using bare and wired probes at two

with the calcium sulphate system. The rate of increase in surface temperature is

directly proportional to the rate of fouling in the batch stirred cell. The results,

, demonstrate that the fouling rate, which is proportional to the gradient, is lower for

. However the mitigating effect of the wires seems

significant as for the case of crude oil fouling. The results also show that the fouling rate is, as

high stirring speed.

Fig. 9 Surface temperature increase with time due to CaSO4 fouling

bare probe, 130 rpm; x: bare probe, 300 rpm

wired probe, 130 rpm; +: wire probe, 300 rpm

Bulk temperature: 55 °C; Initial surface temperature: 85 °

Summary of CaSO4 Scaling Rates on Bare and Wired P

Stirred speed 130 rpm Stirred speed 300 rpm

5.2E-5 m2K/kJ 4.

4.1E-5 m2K/kJ 2.9E

Confidential



heat transfer surface, can be gained more quickly. Fouling experiments were


Bare Probe vs Wired Probe

Fouling rate

(m2K/J)

1.10 x 10-9

Not detected

5.47 x 10-10

°C; Stirring speed: 160 rpm; Crude A

using bare and wired probes at two

The rate of increase in surface temperature is

e results, summarised in Table

nal to the gradient, is lower for the wired

. However the mitigating effect of the wires seems not to be as

show that the fouling rate is, as

fouling

°C

on Bare and Wired Probes

Stirred speed 300 rpm

4.2E-5 m2K/kJ

2.9E-5 m2K/kJ


3.3.2 Influence of surface materials

Table 4 summarises the fouling rate data obtained using

stainless steel. The fouling rate on the probe of stainless steel was

mild steel or copper, even at a significantly higher

found to be similar in level for the mild steel and

Table 4 Summary of Fouling Rate Data Obtained Using Probes of Different M

Probe material and initial surface temperature

Mild steel, 78 °Copper 78 °C

Stainless steel, 90

3.3.3 Asymptotic fouling

The increase in fouling resistance

form (Figure 10). This trend can

provided by Kern and Seaton (1959), as shown in Figure 10 in which the solid line is the Kern and

Seaton model.

Fig. 10 Asymptotic scaling

4. CFD SIMULATION AND MODELLING

CFD simulation and modelling stud

software package, Comsol 4.2. The velocity and temperature distribution in the stirred cell

with a wired probe and in a round tube fitted with a hiTRAN insert have both been

Based on the simulation results,

the average heat transfer coefficient between the wall and fluid in a tube fitted with inserts

obtained. The CFD simulation for

(Yang et al. 2009a). The basic equations

manual 2006a), but are briefly described

momentum are in similar forms to

turbulent dynamic viscosity term to the viscosity expression. The equations of the turbulent kinetic

of 30

Influence of surface materials

Table 4 summarises the fouling rate data obtained using bare probes made of mild steel

The fouling rate on the probe of stainless steel was significantly lower than on either

a significantly higher surface temperature, whilst fou

in level for the mild steel and copper probes.

Table 4 Summary of Fouling Rate Data Obtained Using Probes of Different M

and initial surface temperature Fouling rate

°C 4.33E

C 5.58E

Stainless steel, 90 °C 8.33E

Bulk temperature: 55 °C

fouling resistance with time for the CaSO4 system tends to follow an

form (Figure 10). This trend can be correlated quite well using the simple and original model

(1959), as shown in Figure 10 in which the solid line is the Kern and

Fig. 10 Asymptotic scaling of CaSO4 in the batch stirred cell

SIMULATION AND MODELLING

CFD simulation and modelling studies on fouling have been conducted using a commercial

The velocity and temperature distribution in the stirred cell

a round tube fitted with a hiTRAN insert have both been

the shear stress distribution over the surface of a wired probe,

the average heat transfer coefficient between the wall and fluid in a tube fitted with inserts

The CFD simulation for the stirred cell with a bare probe can be found

The basic equations for the turbulent flow can be found elsewhere (Comsol

described as follows. The equation of continuity and the equation of

momentum are in similar forms to the normal Navier-Stokes equations but with the addition of the


Confidential

made of mild steel, copper, and

significantly lower than on either

surface temperature, whilst fouling rates were

Table 4 Summary of Fouling Rate Data Obtained Using Probes of Different Materials

Fouling rate (m2K/kJ)

4.33E-6

5.58E-6

8.33E-7

to follow an asymptotic

simple and original model

(1959), as shown in Figure 10 in which the solid line is the Kern and

in the batch stirred cell

conducted using a commercial

The velocity and temperature distribution in the stirred cell fitted

a round tube fitted with a hiTRAN insert have both been simulated.

he shear stress distribution over the surface of a wired probe, and

the average heat transfer coefficient between the wall and fluid in a tube fitted with inserts, can be

can be found in previous work

for the turbulent flow can be found elsewhere (Comsol

The equation of continuity and the equation of

Stokes equations but with the addition of the


of 30


energy (k) and the dissipation rate of turbulent energy (ε) are as follows (Comsol, 2006; Wilcox,

2000):

( ) ( )[ ] ρεησ

ηηρ

ρ−∇+∇+

∇

+⋅∇=∇⋅+

∂

∂ 2T

T

k

T uukkut

k (1)

( ) ( )[ ]k

CuukCCut

TT

2

2

2

1

ερρε

σ

ηηερ

ερεµε

ε

−∇+∇+

∇

+⋅∇=∇⋅+

∂

∂ (2)

Here, ηT = ρCµk2/ε is the turbulent dynamic viscosity. Values of the k-ε model parameters are given

in the literature (Comsol 2006). For the heat transfer equations, the turbulence results in an effective

thermal conductivity keff (Comsol, 2006):

keff = ko + kT (3)

kT = CpηT (4)

4.1 CFD SIMULATION OF FLOW IN THE STIRRED CELL WITH A WIRED PROBE

The CFD model geometry is set to be three dimensional. The properties of Crude A have been

given in Table 1, and more details, such as density and viscosity can be found in Yang et al.

(2009a). For the CaSO4 – water system, given the low salt concentration, the physical properties for

water built into the Comsol material library can be used directly.

4.1.1 Effect of shear stress

Figure 11 shows the velocity field for the flow of Crude A in the batch stirred cell fitted with the

wired probe.

Fig. 11 Velocity field in the stirred cell with a wired probe

Fluid: Crude A; Stirrer speed: 200 rpm

Bulk temperature: 523 K


Figure 12 shows the shear stress distribution over the probe surface around a circle, 0

middle height of the probe. It is see

probe due to the interference to the flow by the wires.

probe are also shown in this figure.

is higher than that over the bare probe. This helps to explain

given previously in Tables 2 and

Fig 12 Comparison of shear stress over th


Figure 13 shows the vertical distribution of shear stress in front

probe surface. Given the clockwise flow

than behind it. This may suggest that fouling is more likely behind the wire. A photo

probe after a fouling test confirms this prediction

Figure 13 Vertical distribution

Left: Diagram of appropriate

Red point – wire location; B

Right:

of 30

shows the shear stress distribution over the probe surface around a circle, 0

It is seen that the shear stress varies in a periodical manner around

probe due to the interference to the flow by the wires. For comparison, shear str

lso shown in this figure. It can be seen clearly that the shear stress over the

t over the bare probe. This helps to explain the comparative

3 for crude oil and the calcium sulphate solution, respectively.

Comparison of shear stress over the probe surface – around a circle (0


Bulk temperature: 523 K

shows the vertical distribution of shear stress in front of and behind the wire

. Given the clockwise flowing direction, the shear stress is higher in front of th

. This may suggest that fouling is more likely behind the wire. A photo

confirms this prediction (Figure 14).

distribution of shear stress in front of and behind the wire

Left: Diagram of appropriate locations around the probe

wire location; Blue points – positions where shear stress values are calculated

Arrow: flow direction

Right: Vertical distribution of the shear stress

Confidential

shows the shear stress distribution over the probe surface around a circle, 0 - 2π, at the

periodical manner around the

For comparison, shear stress data for a bare

that the shear stress over the wired probe

comparative fouling rate results

3 for crude oil and the calcium sulphate solution, respectively.

around a circle (0 - 2 π)

and behind the wire over the

tion, the shear stress is higher in front of the wire

. This may suggest that fouling is more likely behind the wire. A photograph of the

and behind the wire

locations around the probe

shear stress values are calculated;

of 30


Fig. 14 Photograph of the probe after CaSO4 fouling test

Fluid flows clockwise when viewed from the top

4.1.2 Effect of temperature

Figure 15 shows the temperature field of the CaSO4 – water fluid in the stirred cell with the wired

probe. Figure 16 shows vertical distributions of the temperature over the positions where the

thermocouples are located, and over the probe surface. The temperature readings given by the

thermocouples were in good agreement with those obtained by the CFD model simulation. This can

be regarded as a good validation of the model. The fouling layer thickness on the bare probe was

measured using a Proscan 2000 machine (Yang et al. 2009a). Figure 17 shows the fouling layer

thickness profile. It can be seen that the fouling layer thickness profile over the probe surface is

strikingly similar in form to the temperature profile shown in figure 16, with their maximum

occurring near to the middle of the heated probe surface.

Fig. 15 Temperature field of CaSO4 fluid in the stirred cell with a wired probe

Bulk temperature 55°C; Average heat flux from the probe surface: 31 kW/m2

Stirrer speed: 130 rpm


Figure 1

Blue line: actual probe surface temperature

line where the thermocouples are

thermocouple twm; ♦: temperature reading by

temperature: 55°C; Average heat flux: 31 kW/m

Fig. 17 Fouling layer

The shear stress over the bare probe surface where the fouling zone is located was relatively

constant, as illustrated in Fig. 12

temperature. Accordingly, the similarity of the temperature and the fouling layer thickness profiles

provide a qualitative explanation for

4.2 CFD SIMULATION FOR HEAT TRANSFER

The CFD simulation was focused on

results of the velocity field and the shear stress distribution have been

Crittenden 2011).

4.2.1 Temperature field in tube fitted with insert

The hiTRAN®

inserts comprise a series of loops equally spaced with a helical pattern and

periodical pattern in the axial dimension. For CFD

round loops whose diameter and thickness are set to be the same as for the actual insert.

300

310

320

330

340

350

360

0

Tem

pera

ture

(K

)

Vertical position from the probe shoulder

of 30

Figure 16 Vertical distribution of temperature

probe surface temperature (by simulation); Red line: temperature over the vertical

are located in the probe (by simulation); ■: temperature reading by

: temperature reading by thermocouple twb. Stirrer speed:

temperature: 55°C; Average heat flux: 31 kW/m2

Fouling layer thickness profile obtained using Proscan 2000

he shear stress over the bare probe surface where the fouling zone is located was relatively

Fig. 12. Hence the fouling behaviour is solely determined by the surface


explanation for the fouling behaviour in the experiment.

HEAT TRANSFER IN TUBE FITTED WITH

FD simulation was focused on heat transfer in tubes fitted with hiTRAN

the shear stress distribution have been reported elsewhe

Temperature field in tube fitted with insert

inserts comprise a series of loops equally spaced with a helical pattern and

periodical pattern in the axial dimension. For CFD simulation, the inserts are represented by closed

round loops whose diameter and thickness are set to be the same as for the actual insert.

0.02 0.04 0.06 0.08Vertical position from the probe shoulder

(m)

Confidential

; Red line: temperature over the vertical

: temperature reading by

speed: 130 rpm; Bulk

obtained using Proscan 2000

he shear stress over the bare probe surface where the fouling zone is located was relatively

. Hence the fouling behaviour is solely determined by the surface


AN INSERT

hiTRAN inserts, whilst the

reported elsewhere (Yang and

inserts comprise a series of loops equally spaced with a helical pattern and a

, the inserts are represented by closed

round loops whose diameter and thickness are set to be the same as for the actual insert. The model


tube is divided into three sections, namely pre

sections as shown in Figure 17. The boundary conditions for all walls of the solid domains/metal

phases in the pre- and post- insert sections are set to be as for thermal insulation. The outer wall in

the insert section is set to be at constant temperature (250°C or 5

thermal wall function. This arrangement would simplify the calculation of the average heat transfer

coefficient using the temperature distributions obtained from the simulation, which will be

described later. Figure 18 shows the resulting temperature field.

Fig. 17 Division of the tube length for simulation purposes

Fig. 18 Temperature field in tube fitted with inserts

inlet linear velocity: 1m/s; bulk temperature: 423K

It is notable that the fluid temperature is higher at the location just behind the wire loop, where the

shear stress is lower according to Yang and Crittenden (2011).

temperature and the film temperature (which is simply an

have both been used in previous fouling research investigations. Nonetheless, little attention has

actually been paid to the effect of the local fluid temperature near to the wall. This local fluid

temperature near to the wall may play an important role in the crude oil fouling process, as it can

have a significant influence on the phase

believed to be a key aspect of the crude oil fouling process (Macchietto e

no experimental results are available to demonstrate the effect of fluid temperature near

on fouling either in the case of bare round tubes or

It is interesting to plot the local surf

Pre-insert section.

Thermal insulation

-0.02 m 0.0 m

of 30

tube is divided into three sections, namely pre-insert (insulated), insert, and post

shown in Figure 17. The boundary conditions for all walls of the solid domains/metal

insert sections are set to be as for thermal insulation. The outer wall in

the insert section is set to be at constant temperature (250°C or 523K), and the inner wall

This arrangement would simplify the calculation of the average heat transfer


s the resulting temperature field.


Temperature field in tube fitted with inserts

inlet linear velocity: 1m/s; bulk temperature: 423K


shear stress is lower according to Yang and Crittenden (2011). The effects of both wall surface

temperature and the film temperature (which is simply an average of wall and bulk temperatures)



wall may play an important role in the crude oil fouling process, as it can

have a significant influence on the phase behaviour of asphaltenes present in the oil, a phenomenon

believed to be a key aspect of the crude oil fouling process (Macchietto et al.,

no experimental results are available to demonstrate the effect of fluid temperature near

on fouling either in the case of bare round tubes or the tubes fitted with inserts.

It is interesting to plot the local surface temperature over the wall as a function of the local shear

Horizontal slice

Vertical slice

Temperature scale (K)

Insert section, Surface

temperature Ts Post-insert section. Thermal

insulation

0.0 m 0.033 m 0.08 m

Confidential

insert (insulated), insert, and post-insert (insulated)

shown in Figure 17. The boundary conditions for all walls of the solid domains/metal

insert sections are set to be as for thermal insulation. The outer wall in

23K), and the inner wall to be a

This arrangement would simplify the calculation of the average heat transfer




The effects of both wall surface

average of wall and bulk temperatures)



wall may play an important role in the crude oil fouling process, as it can

present in the oil, a phenomenon

, 2011). Unfortunately,

no experimental results are available to demonstrate the effect of fluid temperature near to the wall

tubes fitted with inserts.

ace temperature over the wall as a function of the local shear

insert section. Thermal

0.08 m

of 30


stress, and to compare this plot with the fouling threshold conditions reported elsewhere by Yang

and Crittenden (2011). As seen in Figure 19, the local conditions of surface temperature and surface

shear stress fall in the fouling zone for operation at an outer wall temperature of 523K and an

average inlet velocity of 1 m/s. In contrast, the local conditions fall within the non-fouling zone for

the same outer wall temperature but at the higher average velocity of 3.6 m/s. These results are

confirmed by the experimental results (Phillips, 1999) in which fouling did occur at this surface

temperature with an average velocity of 1 m/s for the tube fitted with a mid-density insert. The

experimental results also showed that fouling occurred for the bare tube at this temperature and an

average velocity of 3.6 m/s, though no experimental results were obtained under these conditions

for the tube fitted with the insert. These results indicate that the simulation is indeed able to help in

identifying the appropriate operational conditions to eliminate or reduce fouling by taking into

account the actual local conditions.

Fig. 19 Plot of local surface temperature against local shear stress in the plane of fouling threshold

conditions

♦: inlet velocity 1 m/s; ●: inlet velocity 3.6 m/s

Inlet/Bulk temperature: 423K

Outer wall temperature: 523K

■: threshold conditions (Yang and Crittenden, 2011) converted to shear stress from equivalent

velocity

4.2.2 Average heat transfer coefficient

A simple method has been developed to calculate the heat transfer coefficient based on the

temperature distribution. Assuming a small portion of fluid in an annulus of diameter r, thickness

∆r, and unit height passes a distance L from the bottom to the top as shown in Figure 20, the

amount heat gained by this fluid portion is given by:

)(2 brtrp TTrrcq −∆=∆ πρυ (5)

In this equation Ttr and Tbr are the temperatures at the top (radius r) and bottom (radius r),

respectively, and υ is the linear velocity. The total heat obtained by the fluid contained in a cylinder

of radius R and unit height is therefore given by:

400

450

500

550

600

0 20 40 60

Shear stress (Pa)

Su

rface t

em

pera

ture

(K

)

of 30


∫ −=R

brtrp drTTrcq0

)(2πρυ (6)

Fig. 20 Diagram for calculation of an average heat transfer coefficient

In practice, the calculation is conducted by numerical integration, given that Ttr and Tbr are results

obtained from the model simulation. The heat transfer coefficient can then be calculated as follows:

)(2 bs TTRL

qh

−=

π (7)

Here Ts and Tb are the temperatures at the surface and in the bulk fluid, respectively, which are

assumed to be constant. Figure 21a shows that the temperature distribution over the radius near the

left hand end, that is in the pre-insert section at z = -0.02m, is essentially constant, as expected.

Figure 21b shows the temperature distribution for the same average inlet velocity over the radius

near the right hand end of the insert section, that is at z = 0.08m. It should be noted that at this

location the temperature distribution is almost axially symmetric, with a smooth profile across the

radius, with only 1.6 K difference from the centre to wall.

Table 5 shows a comparison of the average heat transfer coefficients obtained (i) by simulation, (ii)

from the Dittus-Boelter correlation, and (iii) from experiments. The method for calculation of the

heat transfer coefficient based on the experimental measurements of the inlet and outlet

temperatures can be found elsewhere (Phillips, 1999). The Table reveals that the simulated values

of the heat transfer coefficient are in broad agreement with the experimental values. The increase in

∆rr

L

Fig. 21a Temperature distribution over radius in

the pre-insert section (z = -0.02 m)

Outer wall temperature at insert section: 523K

Fig. 21b Temperature distribution over radius near

the end of the post-insert section (z = 0.08 m)

Outer wall temperature at insert section: 523K

of 30


the heat transfer coefficient when an insert is used means that the temperature in the shell side of an

exchanger can be reduced for a given thermal duty, so helping to reduce the fouling potential.

Table 5 Average heat transfer coefficient (W/m2K)

Velocity

(m/s)

h

bare tube

simulation

h

Bare tube

Dittus-Boelter

method

h

Experimental

value

(Phillips

1999)

h

Tube with insert

experimental value

Phillips (1999)

h

Tube with insert

by CFD simulation

0.5 443 391 490 1460 1644

1.0 826 682 780 2150 2292

2.0 1365 1186 1280 3460 3768

4.3 RESEARCH ON FOULING MODELS

4.3.1 Fouling rate and fouling thresholds

The fouling rate in a given heat exchanger is determined by the operational parameters, namely,

velocity and surface temperature. Most established models are for fouling in the tube-and-shell type

heat exchangers, in which the velocity and Reynolds number are well defined. These models are,

accordingly, not straightforward to use in heat exchangers with complicated configurations, for

instance in tubes fitted with inserts or for plate heat exchangers. To extend the application scope of

these fouling models, the concept of equivalent velocity or equivalent Reynolds number has been

introduced (Yang et al. 2009a, Yang and Crittenden 2011). It is defined to be the velocity/Re

number in a bare tube that gives the same wall shear stress in a tube of the same internal diameter

fitted with inserts and operating at a different average fluid velocity. The shear stress and velocity

data can be obtained from the CFD simulation or from empirical correlations, as has been carried

out for plate heat exchangers by INTHEAT project partner SODRU.

A modified version of Yeap’s model (Yeap et al. 2004) has been used to correlate the fouling rate

data of Maya crude oil (Yang and Crittenden 2011):

8.0

3/23/13/123

3/43/23/2

)/exp(1uC

RTETCuB

TuCA

dt

dR

ssf

sff−

+=

−−

−

µρ

µρ (8)

This model, initially developed for crude oil fouling, is capable of taking into account the effects of

both mass transport and chemical reaction in fouling (Yeap et al. 2004), but at the expense of a little

complication. This model has successfully been applied to predict fouling of other fluids in plate

heat exchangers by SODRU. The original Ebert and Panchal (1997) model, shown in equation (9)

has also been used to fit the fouling rate data including both the positive and negative rates shown in

Figure 6 in section 3.2.1:

γτα β −=−

f

A

RT

E

feRe

dt

dR (9)

of 30


Figure 22 shows that the quality of the model fit is reasonably good. Departing from the practice in

Ebert and Panchal, the surface temperature was used in the present study rather than the film

temperature Tf. Also, as explained earlier in this report, the equivalent Reynolds number (Young et

al., 2011), was used in the model. The parameter γ was obtained from the gradient of the plots of

fouling rate versus shear stress shown in Figure 6. It should be recalled that all the gradients can be

considered to be the same and the value was found to be 3.85 m2K/JPa. The remaining parameters

in the Ebert and Panchal model were then obtained by curve fitting. The parameter values that gave

the best fittings were 1190 m2K/J, -0.88 (non-dimensional), and 98.4kJ/mol for α, β, and EA,

respectively. The predictions correspond well with the experimental threshold values, as shown in

Figure 23.

Fig. 22 Comparison of the experimental data and the model fit for the fouling rate

Fig. 23 Comparison of the threshold conditions between the model prediction and experimental

result

▲: experimental result; ■: model prediction

It should be noted that whilst the fouling rate data can be correlated reasonably well by the Ebert

and Panchal model, this model is not necessarily the sole or unique one that is able to interpret the

observed fouling behaviour. Indeed, the experimental fouling rate data can be correlated just as

successfully using other models, such as the modified Yeap model (Yang and Crittenden 2011),

-1.20E-08

-6.00E-09

0.00E+00

6.00E-09

1.20E-08

-1.2E-08 -6.0E-09 0.0E+00 6.0E-09 1.2E-08

Fouling rate - experimental (m2K/J)

Fo

ulin

g r

ate

- m

od

el fi

ttin

g (

m2K

/J)

500

550

600

650

700

0 0.5 1 1.5 2

Shear stress (Pa)

Th

res

ho

ld t

em

pe

ratu

re (

K)

of 30


though details of these models’ fit are not provided. Given its simplicity, a model of this type with

the parameters generated was passed onto the project partner at the University of Manchester.

4.3.2 Compensation plot

It had been initially thought that the compensation plot might provide an opportunity to create a

single modeling method to account for all crude oil fouling (Young et al., 2011), and to account for

all calcium sulphate scaling. Both these types of fouling show strong dependencies on both surface

temperature and fluid velocity (shear stress). The compensation plot stems from the Arrhenius plot

which is the linear relationship between the logarithm of the fouling rate and the reciprocal of the

absolute temperature. Whilst good Arrhenius plots are indeed found in both crude oil fouling and

calcium sulphate scaling, the slopes of the plots are seen to be dependent upon velocity or Reynolds

number. Hence, the apparent activation energy seems to increase with the degree of turbulence.

The compensation plot is a graph of the logarithm of the pre-exponential factor against the apparent

activation energy. Because the compensation plots show very high correlation coefficients it has

always been suspected that the effects for both crude oil fouling and calcium sulphate fouling have

been “false” rather than “true”. Indeed, a recent paper presented at the 2011 Heat Exchanger

Fouling and Cleaning Conference, presented an elegant mathematical reasoning to suggest the

falseness of the compensation plot in fouling applications.

4.3.3 Modeling of fouling induction periods

A mathematical model to correlate fouling resistance data in both the induction period and the

process thereafter was presented at the International Conference on Heat Exchanger Fouling and

Cleaning 2009 (Yang et al. 2009b), and has been further developed since then by the team at the

University of Bath.

Fouling on a heat exchanger surface may be described in the following manner. Firstly, in the

induction period, the active fouling species adhere to the heat transfer surface and gradually cover it

from a fractional coverage of θ = 0 to total coverage at θ = 1. This pre-conditioning layer is very

thin, though not necessarily a single molecular layer and so the increase in fouling resistance Rf is

negligible. Changes in surface roughness are ignored. Secondly, in the fouling period the fouling

layer may start to grow immediately on the covered/pre-conditioned surface when it may be

assumed that the growth rate is proportional to θ. The overall rate of fouling resistance growth can

therefore be expressed as:

′= f

fR

dt

dRθ (10)

Here, Rf′ can be any form of established fouling rate expression, such as those proposed by

Crittenden et al. (1987a and 1987b), Epstein (1994), Ebert and Panchal (1997), etc. These models

describe the fouling rate in a near linear form. In the early stage of surface pre-conditioning, active

species can be captured and adhered to the surface. The following relationship may then apply:

)1( θθ

−∝dt

d (11)

of 30


Meanwhile, the particles that stick to the surface act as seeds, attracting more foulant around them,

such that fouling proceeds in a micro-growth manner. The growth rate is assumed to be first order

in fractional surface coverage θ:

θθ

∝dt

d (12)

Combining the two aspects gives the coverage growth rate as follows:

Growth rate = k1θ(1-θ) (13)

Adopting the concept of removal or release from the surface as in adsorption science, the removal

rate of the surface coverage is assumed to be proportional to the surface coverage. That is:

Removal rate = k2θ (14)

Combining Eqs. (4) and (5) yields the net growth rate:

θθθθ

21 )1( kkdt

d−−= (15)

Integrating Eq. (15) gives:

tkkeck

kk)(

1

21

211

1−−+

−=θ (16)

Given the growth of θ reaching its maximum at 0.5 θmax, the time, t0.5 at which θ reaching 0.5θmax is

defined as the induction time. According to Eq. 16, t0.5 is given by:

21

5.0

ln

kk

ct

−= (17)

In cases of significant fouling, i.e. k1>>k2, the value of k2 in Eq. (17) may be negligible, and k1 is a

lumped growth rate constant which may be assumed to depend on the surface temperature

according to the Arrhenius equation:

RTE

iieAk/

1

−= (18)

Taking logarithms of Eq. (17) gives the temperature dependency of the induction period:

RT

EAct i

i

−−−= ln)ln(lnln 5.0

(19)

It was demonstrated that this model is capable of interpreting the effect of temperature on induction

time for both crude oil and protein fouling (Yang et al. 2009b). Further development of the model

of 30


allowed the effect of velocity to be predicted. Adopting the concept given by Polley et al. (2007)

and Yeap et al. (2004), the removal term is proportional to the 0.8 power of velocity, and hence the

removal parameter k2 in the induction period model may be expressed as follows:

8.0

2 uk γ= (20)

Figures 24 and 25 shows the model fittings of the experimental data by Mwaba et al. (2006), and by

Geddert et al. (2009), respectively, revealing that good fits for the fouling curves can be obtained.

Table 6 summarises the model parameters for the data of Geddert et al. (2009), and shows that the

predicted induction times match those observed experimentally quite well.

Table 6 Model Parameters for Fitting of the Data of Geddert et al. (2009)

Coating Re

k1

(1/hour)

k2

(1/hour)

γ

c

t0.5 (hour)

observed

t0..5 (hour)

model

No coating 1030 2.57 0.33 0.0013 1400 3 3.2

No coating 3010 2.57 0.79 0.0013 1400 4 4.1

SICON 1030 0.66 0.196 7.61E-4 251000 23 26

SICON 3010 0.66 0.462 7.61E-4 251000 65 63

Fig. 24 Effect of velocity - model fittings for the data of Mwaba et al. (2006)

From left to right: 0.3 m/s, 0.6 m/s, 1.0 m/s

Symbols: Mwaba experimental data; Lines: model fittings

Fig. 25 Model fittings for the data of Geddert et al. (2009)

From left to right: Stainless steel Re 1030, Re 3010, SICON coating Re 1030, Re3010

Symbols: Geddert et al. data; Lines: model fittings

0

2

4

6

8

10

12

14

0 2000 4000 6000

Time (min)

Fo

ulin

g r

ate

(x10

-4 m

2K

/W)

-0.2

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 20 40 60 80 100

Time (hour)

Fo

ulin

g r

ate

(*1

0-4

m2K

/W)

of 30


5. CONCLUSION

Fouling is fundamentally a complex phenomenon and a broad range of basic fouling mechanisms

exist, and hence a range of experimental technologies need to be developed for the investigation of

fouling and its mitigation. At Bath, fouling experiments have been carried out over a wide range of

conditions using a batch stirred cell system. Negative fouling rates are observable if the surface

temperature is reduced and/or the stirring speed is increased after the test surface has undergone a

significant amount of fouling, i.e. the fouling resistance has increased to a significant level. The

fouling rate data for both positive and negative fouling rates can then be utilized to identify the

fouling threshold conditions relatively quickly.

It has been demonstrated that CaSO4 crystallisation fouling can be investigated using the stirred cell

system. The results of both crude oil and calcium sulphate fouling have been successfully

interpreted by CFD simulation, including the effects of shear stress and surface temperature on

fouling rates.

With the help of CFD simulation, the concept of equivalent velocity / equivalent Reynolds number

is developed such that a fouling model developed for bare round tubes can be extended for use with

more complex geometries. It has been demonstrated, for example, that the Ebert and Panchal model

(1997) and the Yeap model (Yeap et al. 2004; Yang and Crittenden 2011) can both be adapted

successfully to correlate the fouling data of a crude oil obtained using the batch stirred cell.

Recently this approach has been adopted for modelling fouling in plate heat exchangers by

INTHEAT partner SODRU. Fouling threshold conditions can be predicted successfully, auguring

well for the development of successful strategies to mitigate the highly energy consuming fouling

problem.

CFD simulation has confirmed that the average heat transfer coefficient for a tube fitted with a

hiTRAN insert is much higher than the bare tube operated under the same conditions of surface wall

temperature and average velocity. The CFD simulation can be used to help determine the optimum

design of insert for a particular application, whether fouling or non-fouling. The increase in the heat

transfer coefficient when an insert is used means that the temperature in the shell side of an

exchanger can be reduced for a given thermal duty, so helping to reduce the fouling potential.

The model for fouling induction makes it possible to describe the fouling process from the start of

the induction period up to the steady fouling rate stage using a single and simple mathematical

expression. The induction period model has been demonstrated with applications in crude oil

fouling, calcium sulphate fouling and whey protein fouling. The model is able to describe

quantitatively the influence of surface temperature on the length of the induction period for the

crude oil and whey protein fouling systems, and to describe the influence of velocity on the

induction time in a semi-quantitative manner. The proposed term t0.5 which is the time to reach 50%

of the maximum surface coverage, θmax, provides a practical measure of the length of the induction

period.

6. REFERENCES

Bennett, C.A., Kistler, R. S., Nangia, K., Al-Ghawas, W., Al-Hajji, N. And Al-Jemaz, A., (2009),

Observation of an isokinetic temperature and compensation effect for high-temperature crude oil

fouling, Heat Transfer Engineering, 30, 794-804.

of 30


Bott, T. R., (1990), Fouling Notebook, Institution of Chemical Engineers, Rugby.

Bott, T. R., (1993), Aspects of biofilm formation and destruction, Corrosion Reviews 11 (1-2) 1-24.

Comsol, (2006), Model library - chemical engineering module, pp. 238-240

Crittenden, B. D., Hout, S. A. and Alderman, N. J., (1987a), Model experiments of chemical

reaction fouling, TransIChemE, 65, 165-170.

Crittenden, B.D., Kolaczkowski, S. T. and Hout, S. A., (1987b), Modelling hydrocarbon fouling,

TransIChemE, 65, 171-179.

Crittenden, B. D., Kolaczkowski, S. T., and Downey, I. L., (1992), Fouling of crude oil preheat

Exchangers. TransIChemE, 70, 547-557.

Crittenden, B. D., Kolaczkowski, S. T. and Takemoto T., (1993), Use of in-tube inserts to reduce

fouling from crude oils, AIChE Symp Series, 89 (295), 300-307.

Crittenden, B.D., Kolaczkowski, S. T., Takemoto, T. and Phillips, D. Z., (2009), Crude oil fouling

in a pilot-scale parallel tube apparatus, Heat Transfer Engineering, 30, 777-785.

Eaton, P., (1983), Laboratory Fouling Test Apparatus for Hydrocarbon Feedstocks, US Patent,

4383438, 1983.

Eaton, P. and Lux, R., (1984), Laboratory fouling test apparatus for hydrocarbon feedstocks. In

Fouling in Heat Exchange Equipment, ASME HTD 35, pp 33-42.

Ebert, W., and Panchal, C. B., (1997), Analysis of Exxon crude oil slip stream coking data, In

Fouling Mitigation of Industrial Heat Exchange Equipment, C. B. Panchal and W. C. Kuru (Eds.),

Begell House, New York, pp. 451-460.

Epstein, N., (1981), Fouling in heat exchangers, In Fouling of Heat Transfer Equipment, E.F.C

Somerscales and J. G. Knudsen, J.G. Washington D.C.: Hemisphere Publishing Corporation, pp.

701-734.

Epstein, N., (1994), A model of the initial chemical reaction fouling rate for flow within a heated

tube and its verification, Proc 10th

International Heat Transfer Conference, Institution of Chemical

Engineers, Rugby, Vol. 4, pp. 225-229.

Geddert T., Bialuch I., Augustin W., and Scholl S., Extending the induction period of crystallization

fouling through surface coating, Heat Transfer Engineering, 30: 10-11, (2009) 868-875.

Ishiyama, E. M., Paterson, W. R. and Wilson, D. I., (2009), Platform for techno-economic analysis

of fouling mitigation, Energy & Fuels, 23, 1323-1337.

Kern, D. Q., and R .E. Seaton, (1959), A theoretical analysis of thermal surface fouling, Brit. Chem.

Eng., 4, 258-262.

of 30


Knudsen, J. G., Dahcheng, L. and Ebert, W. A., (1997), The determination of the threshold fouling

curve for a crude oil, In Understanding Heat Exchanger Fouling and its Mitigation, T.R. Bott, L.F.

Melo, E. F. C. Somerscales, C. B. Panchal, (Eds.), Begell House, pp. 265-272.

Krueger, A. W, and Pouponnot, F., Heat exchanger performance enhancement through the use of

tube inserts in refineries and chemical plants – successful application examples: Spirelf, Turbotal

and Fixotal systems. Proc. Eurotherm Conference on Fouling and Cleaning in Heat Exchangers,

Schladming, Austria, 2009, pp. 400-406.

Macchietto, S., Hewitt, G. F., Coletti, F. F., Crittenden, B. D., Dugwell, D. R., Galindo, A.,

Jackson, G., Kandiyoti, R., Kazarian, S. G., Luckham, P. F., Matar, O. K., Millan-Agorio, M.,

Müller, A., Paterson, W., Pugh, S. J., Richardson, S. M. and Wilson, D. I., (2011), Fouling in crude

oil preheat trains: a systematic solution to an old problem, (2011), Heat Transfer Engineering, 32,

197–215.

Melo, L. F., Bott, T. R. and Bernardo, C. A. (1987), Fouling Science and Technology, Kluwer

Academic Publishers, Dordrecht.

Mwaba, M. G. , Rindt, C. C. M. , Van Steenhoven, A. A. and Vorstman, M. A. G. Experimental

Investigation of CaSO4 Crystallization on a Flat Plate, Heat Transfer Engineering, 27: 3, (2006), 42

– 54.

Panchal, C. B., Kuru, W. C., Liao, C. F., Ebert, W. A. and Palen, J. W., (1997), Threshold

conditions for crude oil fouling, In Understanding Heat Exchanger Fouling and its Mitigation, T. R.

Bott, L. F. Melo, C. B. Panchal and E.F.C. Somerscales, (Eds.), Begell House, pp. 263-281.

Phillips, D. Z., (1999), Mitigation of crude oil fouling by the use of HiTRAN inserts, PhD Thesis,

University of Bath.

Polley, G.T., D.I. Wilson, D. I., Pugh, S. J. and Petitjean, E., (2007), Extraction of crude oil fouling

model parameters from plant exchanger monitoring, Heat Transfer Engineering, 28, 185-192.

Ritchie, J. m. and Droegemueller, P., Application of tube inserts in heat exchangers: Benefits of tube

inserts, in Heat exchanger design handbook, ed. G. F. Hewitt, Begell House, Redding, CT, Section 3.21.2.

(2008).

Takemoto, T., Crittenden, B. D. and Kolaczkowski, S. T., (1999), Interpretation of fouling data in

industrial shell and tube heat exchangers, TransIChemE, 77, 769-778.

Watkinson, A. P. and Wilson, D. I., (1997), Chemical reaction fouling: a review, Experimental

Thermal and Fluid Science, 14, 361-374.

Wilcox, D. C., (2000), Turbulence modelling for CFD, 2nd

edition, DCW Industries Inc, La Canada,

CA, pp 123-127.

Wills, A., Bott, T. R. and Gibbard, I. J., (2000), The control of biofilms in tubes using wire-wound

inserts, Canadian J. Chem. Eng., 78, 61-64.

of 30


Yang, M., A. Young, and B.D. Crittenden, (2009a), Use of CFD to correlate crude oil fouling

against surface temperature and surface shear stress in a stirred fouling apparatus, Proc. Conf. Heat

Exchanger Fouling and Cleaning VIII, Schladming, Austria: www.heatexchanger-fouling.com, pp.

272-280.

Yang, M., Young, A., Niyetkaliyev, A., and Crittenden, B. D., (2009b), Modelling of fouling

induction period. Proc. Conf. Heat Exchanger Fouling and Cleaning VIII., Schladming, Austria:

www.heatexchanger-fouling.com, 2009, pp. 69-75.

Yang, M. and Crittenden, B. D., (2011), Fouling thresholds in bare tubes and tubes fitted with

inserts, Applied Energy, in press.

Yeap, B. L., Wilson, D. I., Polley, G. T., and Pugh, S. J., (2004), Mitigation of crude oil refinery

heat exchanger fouling through retrofits based on thermo-hydraulic fouling models.” TransIChemE,

82, 53-71.

Young, A., Venditti, S., Berrueco, C., Yang, M., Waters, A., Davies, H., Hill, S., Millan, M. and

Crittenden, B. D., (2011), Characterisation of crude oils and their fouling deposits using a batch

stirred cell system, Heat Transfer Engineering, 32, 216-227.

fp7-sme-2010-1 262205/ intheat page 1 of...

Documents