NACE Paper No.

MECCOCT18-12493

The Use of Computational Fluid Dynamics (CFD) to

Resolve Corrosion Problems in Refinery Process Units

M.I.Suleiman, T. Firmansyah, A. Al Maqtari, M. Rakib, M. Al Musharfy

ADNOC Refining Research Centre Division (RCD)
Abu Dhabi National Oil Company (ADNOC)
PO Box 3593, Abu Dhabi
United Arab Emirates

ABSTRACT

In refinery units several corrosion mechanisms are observed in different parts of the process streams.
The forms of these corrosion events and their locations depend on feed composition, and process units
operating parameters such as flow, pressure, temperature, partial pressure of hydrogen (H 2), and
hydrogen sulfide (H2S), chloride content, etc.

One of the chronic corrosion problems reported is the premature failure of reactor effluent exchangers
of hydrotreating, and hydrocracking units; in particular the stripper feed and bottom product streams
exchangers. The nature of the corrosion has been always severe thinning of the tubes outer surface
and sometimes localized pitting and perforations. Corrosion is usually observed in one specific zone of
the exchanger tube bundle, viz. areas between 4 to 8 o’clock shell outlet side, downstream of the last
shell baffle and before the tube sheet. This paper presents one case corrosion history investigation
using CFD simulations to study flow patterns inside the shell side of exchanger. The results have
indicated that the corrosion inherent area had near zero velocity resulting in fluid stagnation. The low
velocity allows the suspended solids and salt-saturated water droplet to settle and form a thin layer of
hard deposit under which severe corrosion ensues. By changing fluid velocity and baffle opening
orientation the CFD simulation was able to indicate the best condition for minimizing stagnation and
therefore mitigating the under-deposit corrosion.

Key words: Computational Fluid Dynamics, Corrosion, Heat Exchangers.

1
 INTRODUCTION

Ammonium salt deposition and under deposit corrosion is a source of corrosion failure in
Hydroprocessing reactor effluent air coolers. 1,2 The increasing amount of nitrogen, chloride and sulfur
in crude oil leads to more NH4CI and NH4HS salts, which then generates deposition on the pipeline wall
under the influence of flow and diffusion. 1

Whilst modifying metallurgy is desirable it usually has high cost impact. So, for a more cost-effective
option, strategies of dealing with the chlorides and other process precursors are more often considered.
3
As the mechanism of corrosion and its rate of attack depend on the nature of environment in which
the corrosion takes place, it is necessary to look at the surrounding fluid flow that could be one of
contributory factors to the ensuing corrosion.

Use of Computational Fluid Dynamics (CFD) simulations as engineering analysis tool in fluid flow and
thermal distribution study has recently gained big attention in many industries including oil and gas.
Many CFD studies on the impacts of flow and thermal distribution to the occurrence of erosion and
corrosion have already been published. 1,4-7 Most notable Zhu et al. 1 discussed the application of CFD
to investigate the crystal and deposition behavior of ammonium chloride using multiphase flow
simulation. Results from their CFD study showed distribution of NH4Cl salt granules at the inlet of
reactor effluent air cooler tube and most importantly, 3-D visual mapping of concentration of solids
deposition, which in turn indicated the location of under deposit corrosion. So, CFD has shown its
benefit and advantage as troubleshooting analysis tool.

This paper discusses the premature failure of reactor effluent exchangers of hydrotreating, and
hydrocracking units; in particular the stripper feed and bottom product streams exchangers. The tube
bundle is U-type and metallurgy of tube is carbon steel conforming to ASTM A-179. 8† The failure is due
to external corrosion (from the shell side) of the tube caused by under deposit corrosion. It is
concentrated at the bottom rows of the tubes near the channel side tube sheet around 6 o'clock
position of the exchanger shell and downstream of the last baffle before the tube sheet. Leaks were
noticed in 30 tubes. A photograph depicting a sample of a failed tube is shown in Figure 1.

Figure 1: Photograph of the failed tubes

Laboratory analysis on the sludge from the heat exchanger as well as on the water and naphtha
streams showed the presence of chlorides, sulfur and ammonia. Moreover, the corrosion deposits were
spot analyzed using Energy Dispersive Spectroscopy (EDS). EDS showed that a notable amount of
chlorides was present beneath the corrosion deposits which created a localized under deposit
corrosion.

†
American Society for Testing and Material (ASTM), 100 Barr Harbor Drive, West Conshohocken, PA
19428-2959, USA

2
As part of troubleshooting strategies, CFD was employed to study the flow patterns across the shell
side of the Heat Exchanger, in particular near the region where the corrosion occurred. CFD showed
full insights of flow pattern inside the shell side of the Heat Exchanger such as regions of stagnant, i.e.
close to zero velocities that may induce solids deposition and thus under deposit corrosion.

The CFD model covers the shell geometries including all the internal baffles and also the impingement
plate downstream of the inlet, as shown in Figure 2. For computational efficiency the tube side of the
Heat Exchanger was not modeled explicitly. However, the bulk temperature profile of the shell was
captured using the results of process simulation (ASPEN Exchanger Design and Rating) and
implemented in the CFD model through User Defined Function (UDF).

Outlet

Inlet

Figure 2: CFD geometry of heat exchanger

GOVERNING FLOW EQUATION

The fluid flow behavior is governed by conservation of mass, momentum and energy. The equation for
conservation of mass, or continuity equation, can be written as follows:

𝜕𝜌
+ ∇. (𝜌𝑣⃗) = 𝑆𝑚 (1)
𝜕𝑡

Where ρ density, t time and v

⃗⃗ velocity vector. The source Sm is the mass added to the continuous
phase from the dispersed second phase (e.g. liquid droplets, solid particles, etc.) and any other user-
defined sources.

Conservation of momentum in inertial reference frame is described by

𝜕
𝜕𝑡
(𝜌𝑣⃗) + ∇. (𝜌𝑣⃗𝑣⃗) = −∇. (𝜏̿) + 𝜌𝑔⃗ + 𝐹⃗ (2)

Where p is the static pressure, τ̿ the stress tensor, and ρ ⃗⃗ and ⃗F⃗ are the gravitational body force and
⃗⃗g
external body forces (e.g. that arise from interaction with the dispersed phase), respectively.

3
 CFD SIMULATION CASES AND BOUNDARY CONDITIONS

Initially, a baseline simulation was carried out using exactly the existing condition of baffle arrangement
with the normal operating flow, i.e. 60% of design flow rate. The design flow rate was 93 kg/s. After
evaluating the results of baseline simulations further four CFD simulations were performed with
modifications of baffle geometry and flow rate. Table 1 lists all the CFD simulation cases completed for
this study.

Table 1: List of CFD Simulation Cases

Flow Rate
No Case Geometry
(% of Design Flow)
1 Baseline Existing 60
2 Modification #1 Baffle with One Hole 60
3 Modification #2 Baffle with Four Holes 60
4 Modification #3 Baffle with Full Bottom Cut-out 60
5 Modification #4 Baffle with Full Bottom Cut-out 100
Figure 3 shows baffle modifications with one and four holes consecutively, each having 1” radius.
Figure 4 displays a modified baffle, which is cut out at 120 mm above the base.

Figure 3: Modified baffle with one or four holes at the bottom side

Figure 4: Modified baffle with full bottom cut-out

Fluid properties with temperature variation were taken from ASPEN Exchanger Design and Rating
(EDR), as listed in Table 2 and implemented in CFD as piecewise linear relationship with temperature.

4
 Table 2: List of Fluid Properties

Temperature Pressure Density Specific heat Thermal Viscosity

3
(°C) (bar) (kg/m ) (kJ/kg.K) Conductivity (mPa.s)
(W/mK)
61 13.6 695.8 2.147 0.1145 0.328
71 13.5 686.6 2.197 0.1121 0.297
81 13.5 677.2 2.248 0.1095 0.271
90 13.5 667.9 2.298 0.107 0.248
100 13.5 658.7 2.348 0.1045 0.228
109 13.4 649.4 2.397 0.1019 0.211
118 13.4 640.2 2.446 0.0994 0.195
127 13.4 631.0 2.496 0.0969 0.182
135 13.3 621.8 2.545 0.0943 0.170
144 13.3 612.6 2.595 0.0918 0.159
152 13.3 603.4 2.645 0.0892 0.149
160 13.3 594.2 2.696 0.0865 0.141

The temperature variations across the shell were also calculated in ASPEN EDR and implemented in
the ANSYS FLUENT® ‡ CFD software as a User Defined Function (UDF). The bulk shell temperature
vs. distance is shown in Figure 5.

Figure 5: Shell temperature vs distance in CFD simulations

‡
ANSYS, Canonsburg, Pennsylvania, United States

5
 RESULTS AND DISCUSSION

Figure 6 shows the resulting velocity contours of baseline simulation with red color representing regions
with velocity of 0.1 m/s and above and dark blue regions with velocity close to zero, i.e. stagnant
regions. The velocity contours are displayed in four different planes with three different levels of
horizontal planes and one vertical plane to visually aid the spatial distribution of velocities. Clearly big
stagnant zones appeared at the right side of the fluid inlet, which is expected since it is enclosed by the
tube plate and majority of the fluid flows in the opposite direction towards the outlet. Another dominant
stagnant zone was near the base of the shell opposite of the outlet. This is the area where under
deposit corrosion occurred and led to the tubes leakage.

Tube Plate

Stagnant Zones

Figure 6: Resulting velocity contours of the existing geometry and operation condition

The CFD results have thus indicated that the corrosion inherent area had near zero velocity resulting in
fluid stagnation. The low velocity allows the suspended solids and salt-saturated water droplet to settle
and form a thin layer of hard deposit under which severe corrosion ensues.

Four proposed modifications in terms of baffle geometry and flow rate were conducted to help remove
this area of stagnation. For clarity and comparison purposes, three CFD post-processing vertical
planes, as listed in Table 3, are used. The sketch of these plane positions is shown in Figure 7.

Table 3: Planes used for CFD Post-processing Images

Plane # Position from the Modified Baffle Remarks

(meters)
1 0.02 Near baffle
2 0.2 Cutting the Outlet Tube
3 0.5 Near wall-end

The first modification case was by inserting one hole of 1” radius at the base of the baffle near the outlet.
The resulting velocity contours are shown and compared with the baseline case in Figure 8. The baseline/
original case results are shown at the top whilst the modification ones at the bottom. The left images are at

6
the plane #1 near baffle, the central ones represent those at the plane #2 whilst the right images at the plane
#3 near the wall-end. Clearly the effect of the modified baffle with single hole was very much localized and
still significantly wide areas of stagnation were present at the base of the shell. The same trend of results
was found with the second proposed modification involving four holes at the base of the baffle near the
outlet, as depicted in Figure 9.

Plane #2

Plane #3 Plane #1

Wall-end

Figure 7: Positions of post-processing CFD planes

Original Original Original

Plane 2
Plane 1

Plane 3

Modified One Hole Baffle Modified One Hole Baffle Modified One Hole Baffle

Localized
Effect

Figure 8: Impacts of the modified baffle with single bottom hole

7
 Original Original Original

Plane 1

Plane 2

Plane 3
Modified Four Holes Baffle Modified Four Holes Baffle Modified Four Holes Baffle

Figure 9: Impacts of the modified baffle with four bottom holes

More significant impacts in velocity increase were obtained when the whole bottom section of the baffle
was cut out at 120 mm in height. Figure 10 shows the resulting velocity contours of this modification at
60% of design flow rate. Much wider areas of increased velocity can be seen as a result of this new
opening at the bottom of the baffle. However, some areas with near zero velocities still remain closer to
the far corner. Further removal of stagnant zones can be achieved by increasing the inlet flow rate to
full flow capacity, as displayed in Figure 11. This is the most optimum modification in terms of velocity
increase to remove the areas of solids deposition.

Original Original Original

Plane 1

Plane 2

Plane 3

Modified Full Bottom Opening Modified Full Bottom Opening Modified Full Bottom
Opening
60% Flow

60% Flow

60% Flow

Much Wider
Impact

Figure 10: Impacts of the modified baffle with full bottom opening and 60% design flow

8
 Original Original Original

Plane 1

Plane 2

Plane 3
Modified Full Bottom Opening Modified Full Bottom Opening Modified Full Bottom
Opening
Full Flow

Full Flow
Full Flow

Much Wider
Impact

Figure 11: Impacts of the modified baffle with full bottom opening and full flow

CONCLUSIONS

Computational Fluid Dynamics (CFD) was employed in this study to look into the underlying flow
patterns across the shell side of the Heat Exchanger, in particular near the region where the corrosion
occurred.

Four CFD simulation cases were performed including the existing geometry of the shell side of the Heat
Exchanger and three modifications involving changes in baffle geometry and fluid flow rates. Results
from CFD showed that predominant stagnant zone was present near the base of the original shell,
opposite of the outlet where the tubes leakage occurred. This indicated that the corrosion inherent area
had near zero velocity resulting in fluid stagnation. The low velocity allows the suspended solids and
salt-saturated water droplet to settle and form a thin layer of hard deposit under which severe corrosion
ensues.

By changing fluid velocity and baffle opening orientation the CFD simulation was able to indicate the
best condition for minimizing stagnation and therefore mitigating the under-deposit corrosion.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the support of ADNOC Refining Research Centre Division (RCD)
and Abu Dhabi National Oil Company (ADNOC) Refining management.

9
 REFERENCES

1. M. Zhu, G. Ou, H. Jin, K. Wang, Z. Zheng, “Top of the REAC Tube Corrosion Induced by Under
Deposit Corrosion of Ammonium Chloride and Erosion Corrosion,” Engineering Failure Analysis, 57
(2015), pp. 483-489.
2. P.P. Alvisi, V.F. Cunha Lins, “Acid Salt Corrosion in a Hydrotreatment Plant of a Petroleum
Refinery,” Engineering Failure Analysis, 15 (2008), pp. 1035-1041.
3. National Petroleum Refiners Association (NPRA), “NPRA Question and Answer Session on
Refining and Petrochemical Technology,” (1997), p. 99.
4. S. Nesic, “Using Computational Fluid Dynamics in Combating Erosion-Corrosion,” Chemical
Engineering Science, 61 (2006), pp. 4086-4097.
5. X. Chen, B.S. McLaury, S.A. Shirazi, “Application and Experimental Validation of a
Computational Fluid Dynamics (CFD)-based Erosion Prediction Model in Elbows and Plugged Tees,”
Computers & Fluids, 33 (2004), pp. 1251-1272.
6. M.H. Wang, C. Huang, K. Nandakumar, P. Minev, J. Luo, S. Chiovelli, “Computational Fluid
Dynamics Modeling and Experimental Study of Erosion in Slurry Jet Flows,” International Journal of
Computational Fluid Dynamics, 23 (2009), pp. 155-172.
7. A. Jatale, M. Srinivasa, “CFD Modeling of Fouling in Crude Oil Refinery Heat Exchangers,”
AIChe Spring Meeting, 11th Global Congress on Process Safety, 2015.
8. ASTM A179 / A179M-90a(2012), Standard Specification for Seamless Cold-Drawn Low-Carbon
Steel Heat-Exchanger and Condenser Tubes, ASTM International, West Conshohocken, PA,
2012, www.astm.org

10