Optimising nitrogen utilisation in

refinery operations
Technical aspects and insights on managing nitrogen by considering actual
operational scenarios

Rajib Talukder and Prabhas K Mandal

Aramco

N
itrogen gas, known for being chemically inert, To correctly estimate the flow rate of the blanket gas, it
non-flammable, colourless, odourless, and slightly is important to keep the pressure inside the vessel within
lighter than air at atmospheric temperatures, has safe and operational limits. Here is a basic guide on how to
been extensively used in the chemical and process indus- measure the blanket gas flow rate:
try for many years. The oxygen content in empty vessels, • Assess volume changes due to liquid outflow: The initial
equipment, and pipeline spaces is reduced by it to facilitate step involves identifying the maximum liquid volume that
start-ups, or hydrocarbons are driven out by it on shut- will exit the vessel over a certain period. This shift in volume
downs. Due to its non-reactive nature and low solubility in results in a change in pressure inside the vessel, necessitat-
liquids, it is commonly chosen as a blanketing gas, thereby ing compensation via the blanket gas to maintain positive
virtually eliminating any risk of product contamination. pressure. The Ideal Gas Law is used to convert the change
In refineries, most of the nitrogen is consumed in gaseous in liquid volume into the needed gas volume.
form. Liquid nitrogen is stored and then vapourised into gas- • Consider pressure changes due to liquid contrac-
eous nitrogen as needed. The base nitrogen demand load tion: The impact of major pressure changes in the vessel
for a refinery during normal operation is met by gaseous resulting from variations in incoming feed temperature is
nitrogen, a demand that is an order of magnitude smaller also accounted for, such as the volume contraction caused
than the peak load observed during plant shutdowns. The by the entry of cold feed when the supply of hot feed is
peak load demand is met by liquid nitrogen. interrupted.
This article does not address the methods by which the When calculating the total need for blanketing gas for
overall base or peak nitrogen demands of a refinery are any surge drum, thermal inbreathing is generally not con-
determined. Instead, attention is given to several major sidered along with normal inbreathing resulting from liquid
contributors to the base or peak load where significant opti- movement from the vessels, unlike in the case of tanks. This
misation opportunities are believed to exist due to a high is primarily because the empty vapour space in a surge
degree of conservatism associated with these contributors. drum is much smaller than in tanks, and the likelihood of
In the following sections, major contributors, such as the a coinciding loss of liquid inflow and contraction of the
nitrogen required for tank blanketing, surge drum blanket- vessel’s vapour due to sudden ambient cooling from rain is
ing, and starting up hydroprocessing units, are addressed. extremely low.
Nitrogen blanketing is utilised in vessels/tanks containing However, it is essential to note that nitrogen blanketing
liquids, such as a surge drum or tanks, for the following reasons: of vessels is not designed to address certain circumstances:
• Safety: The use of nitrogen lessens the chance of oxygen • Using blanket gas as a safeguard for the vessel design
penetration, thereby disrupting the formation of the fire for full vacuum: A total feed failure to the surge drum during
triangle (fuel, heat, oxygen). This is especially relevant for the vessel’s emptying process is considered a severe situa-
vessels containing flammable liquid hydrocarbons. tion. In such cases, according to API 521 standards, blanket
• Protection against oxidation: Preventing oxygen from gas cannot be used as a safety measure since control action
entering hinders the oxidation process, which could otherwise credits (for instance, the blanket gas control valve) are not
harm the quality of the liquid inside the vessel. This is particu- acknowledged as safety precautions. The vessel should be
larly important for lean amine and wash water surge drums. either constructed for full vacuum to ensure its safety or
• Prevention of vapour loss: A nitrogen barrier restricts the equipped with a safety instrumentation system to avert a
amount of hydrocarbon vapour that can leave the vessel. vacuum under these emergency circumstances.
• Supporting NPSHa for the pump connected to the surge
Estimation of nitrogen blanketing for surge drums drum: While estimating NPSHa, the blanket gas’s solubility
By maintaining an inert atmosphere over the stored liquid, in the liquids being pumped is considered, and the vapour
nitrogen blanketing ensures quality, regulates pressure, pressure is presumed to be higher than the actual vapour
and prevents incidents. pressure, equal to the surge drum’s normal operating

February 2024 1
[Link]
 requirements for the process units, sub-
Required blanketing nitrogen gas sequently driving up the total nitrogen
demand for the facility. Moreover, it could
Project Project A Project B
Equipment name Lean amine surge drum Lean amine drum result in an overly conservative sizing of
Vessel pressure (bara) 3 2 the nitrogen blanketing valve, which might
Vessel temperature (°C) 70 60 introduce control challenges when the
Vessel head type 2:1 Semi-ellipsoidal 2:1 Semi-ellipsoidal valve output is minimal, possibly triggering
Vessel ID (mm) 3,000 2,500 unnecessary nitrogen wastage. Refiners
Vessel T-T (mm) 10,000 9,000
are adding gap control for the blanketing
Out flow (m³/h) 180 85
control valves, as described later in detail,
Normal liquid level- NLL (mm) 7,000 6,000
Lo Lo liquid level- LALL (mm) 750 500 and this practice results in near zero open-
Nitrogen blanket gas flow (Nm³/h) ing of the nitrogen blanket control valve
1. Method 1 358 139 during normal plant operation.
2. Method 2 63 51 Observations from actual plant data
3. Design value 360 50 reveal a consistent level during normal
operations. During emergency situations
Table 1 when the inflow to the surge drum is dimin-
ished or completely lost, operator actions
pressure. This is done to account for slight degassing in the ensure the feed surge drum level is maintained, preventing
liquid and to prevent pump cavitation. the connected pump from being tripped due to the activa-
tion of the low liquid level trip. The surge drum’s hold-up
Calculation of volume changes due to liquid outflow time is generally set between 10-15 minutes, providing
The volume changes due to liquid outflow can be estimated ample time for operator intervention.
using two different methods: Considering the ample time available for operator inter-
• Method 1: The flow of nitrogen blanketing is calculated vention and the lack of benefits from designing the nitrogen
to maintain the normal operating pressure in the surge blanket flow with high conservativism using Method 1, it is
drum when the outflow from the drum is continuous, but more pragmatic to adopt Method 2 for estimating the nitro-
the inflow to the drum ceases. This estimation uses the API gen blanket gas flow.
2000 liquid outflow method. Licensor does not normally
include thermal inbreathing. Pressure changes due to liquid contraction
• Method 2: The nitrogen blanketing flow is calculated Volumetric contraction in the surge drum can happen due
to maintain the vessel at a slightly positive pressure (1.1 to the replacement of hot feed with cold feed, especially for
bara) when the outflow from the drum is continuous, but hot hydrocarbon feed.
the inflow to the drum fails. This is calculated based on the Reduction of pressure can be estimated as follows:
vapour volume change due to the decrease from the normal
liquid level to the very low liquid level at which the con- Liquid volume change (dV) = VNLL x (1- ϱhot/ϱcold)
nected pump is stopped. Vapour volume above NLL (Vvapour) = VTotal - VNLL
The operation of the surge drum is different from that of
tanks. While tanks are either in receiving or dispatch mode, Pfinal = Pnormal x [Vvapour/( Vvapour + dV)]
surge drums are always in both receiving and dispatch mode.
They are frequently positioned between process units to help Where:
mitigate the impact of flow rate variations between intercon- VNLL = Normal liquid level volume (m³)
nected process units. Unlike the typical control objective of VTotal = Vessel total volume (m³)
maintaining a measurement at a set point, the goal of surge ϱhot = Hot liquid density (kg/m³)
drum level control is to buffer the changes in controlled flow ϱcold = Cold liquid density (kg/m³)
while keeping the liquid level in the vessel within limits. For Pfinal = Final vessel pressure (bara)
surge drums, it is usually more important to allow levels to Pnormal = Vessel normal pressure (bara)
‘float’ to minimise flow rate variations. Therefore, the level r = Density (Rho)
controller must permit this movement and try not to hold the
level close to its set point. Instead, the controller should keep Example calculation:
the surge vessel’s level between its upper and lower limits This example calculation is performed for a diesel hydro-
with the least possible change to its flow output. treater feed surge drum (FSD) when hot feed is replaced
The estimation of nitrogen blanketing for the lean amine with cold feed during hot feed failure. Assuming the FSD
surge drum of two different licensors' diesel hydrotreater will continue to maintain a normal liquid level after replace-
units using both methods is presented in the Table 1. The ment, cold feed with a higher density will have a lower liq-
calculation details can be found in Appendix 1. uid head. The reduction of the liquid head will result in a
Employing Method 1 to determine the nitrogen blanket reduction in liquid volume at a normal liquid level, and this
for the surge drum may lead to a high degree of conserv- volume reduction is estimated considering liquid volumetric
ativism. This strategy could increase the normal nitrogen contraction.

2 February 2024 [Link]

 The hot feed, initially at 125°C, is preheated to 218°C In the case of determining the nitrogen blanketing flow
before entering the FSD. In situations where there is a fail- rate for tanks, the API 2000 method serves as a standard-
ure in the hot feed supply, a cold feed is used as a replace- ised procedure.
ment. This cold feed, starting at 40°C, is then preheated to
155°C before being introduced to the FSD. Estimation of inbreathing due to liquid transfer effect
It is assumed that the FSD has a similar dimension to the According to Section [Link].2 ‘Inbreathing’ of API 2000,
lean amine surge drum of Project A above and the FSD is the estimation of inbreathing, resulting from the maximum
operating at the same pressure of 2.5 bara. From this, the outflow of liquid from a tank, necessitates consideration of
following is calculated: the rate of volume change of tank vapour space due to liq-
uid movement.
VTotal = 78 m³ The requirement for inbreathing, denoted by VinL and
VNLL = 53 m³. measured in normal cubic meters per hour of air, should
ϱ
hot
= 700 kg/m³ equal the maximum liquid discharging capacity for the tank,
ϱ
cold
= 751 kg/m³ denoted by Vliq and measured in cubic meters per hour. This
Pnormal = 2.5 bara correlation is presented in Equation 1. Here, Vliq signifies
Pfinal = To estimate (bara) the rated capacity of the pump connected to the tank:

Liquid volume change (dV) = 53 X (1- 700/751) = 3.6 m³ VinL = Vliq Eq. 1

Vapour volume above NLL (Vvapour) = 78 – 53 = 25 m³ Estimation of inbreathing due to thermal effect
The API-2000 standard, specifically Section [Link].3
Pfinal = 2.5 X (25/(25+3.6) = 2.2 bara
‘Thermal Inbreathing,’ provides guidelines for calculating
From the above calculation, it is evident that the pressure inbreathing attributed to thermal effects. As per this
reduction due to liquid volumetric contraction when replac- section, the thermal inbreathing of the tank, measured in
ing hot feed with cold feed is relatively minor. normal cubic meters per hour of air, is calculated in line
with Equation 2:
Determination of nitrogen requirement for tank
blanketing VinT = C X Vtk0.7 X Ri Eq. 2
Inert gas systems, like those using nitrogen, help prevent
air from entering a tank when there is a chance of vacuum Where:
generation inside it. These systems make tanks safer by VinT is inbreathing flow rate (Nm³/h)
reducing the chance of creating explosive atmospheres and C is a factor that depends on vapour pressure, average
minimising the risk of dangerous flashbacks. storage temperature, and latitude (ref API 2000 Table 2)
However, it is important to note that this nitrogen blan- Vtk, which represents the tank’s vapour volume, is expressed
keting system should not replace vacuum relief devices. in cubic meters. For vertical cylindrical tanks, it is accept-
Despite having an inert gas system in place, the vacuum able to calculate this volume based on the tank shell height,
relief devices need to be large enough to handle situations not including the tank roof.
where the inert gas might not be available.
Vacuum conditions in a tank can occur due to two main The term Ri refers to the reduction factor for insulation.
causes: In situations where no insulation is used, such as this, Ri is
 Liquid transfer effect: This phenomenon occurs when assigned a value of 1.
there is an outflow of liquid from the tank without a corre- When establishing inbreathing requirements, the design
sponding inflow, creating a vacuum. basis must account for the most significant single contin-
 Thermal effect: Changes in atmospheric conditions, gency or any plausible combination of contingencies. To
such as a drop in temperature or shifts in weather patterns ensure comprehensive coverage, the total normal inbreath-
(like wind changes or precipitation), can lead to the con- ing of the tank should, at a minimum, consider the combined
traction or condensation of vapours, consequently resulting effects of liquid transfer and thermal conditions during
in a vacuum. normal operations. As a result, the normal inbreathing

C factors
Latitude C factor for various conditions
Vapour pressure similar to hexane Vapour pressure higher than hexane, or unknown
Average storage temperature, ºC
<25 ≥ 25 <25 ≥ 25
Below 42° 4 6.5 6.5 6.5
Between 42° and 58° 3 5 5 5
Above 58° 2.5 4 4 4

Table 2

February 2024 3
[Link]
requirement for the tank is derived by summing the values pragmatic approach, a more realistic estimation of nitrogen
from Equations 1 and 2, capturing both the liquid transfer demand for tank inbreathing can be achieved, potentially
and thermal influences on the inbreathing system. reducing the overall nitrogen consumption significantly.
Frequently, when estimating the total nitrogen inbreath-
Alternate method for estimating nitrogen ing requirement for a tank farm, a common practice is to
inbreathing quantity add the inbreathing due to liquid movement to the thermal
According to API 2000 section 3.5.3, when an inert gas inbreathing for all tanks. However, it is important to note
system is employed to prevent the entry of air into the that, for a specific service, typically only one tank operates
tank during vacuum conditions, thereby reducing the risk in despatch mode, driven by the connected pump. This gen-
of a potentially explosive atmosphere inside the tank, the eral addition of inbreathing due to liquid movement leads
Annex F method can be utilised to estimate inert gas blan- to a significant increase in the normal nitrogen inbreathing
keting for tanks. requirement for the entire tank farm. As a result, a more
In the context of refineries, tank vents are typically with- accurate and pragmatic approach is needed to optimise
out any flame arrester, and thus, the sizing of venting nitrogen usage and better reflect the actual operational
devices is determined using Annex F Level 3 equation, as conditions of the tanks.
shown below: Various interpretations and methodologies have been
observed when estimating nitrogen inbreathing for a tank
VI = 0.5C • RiVtk + Vpe
0.7
farm. To facilitate a comparative analysis of inbreathing
flow rates using different methods, nitrogen blanketing for
VI = 0.12 • Vtk a selection of representative tanks with representative out-
flow has been estimated using the following approaches:
Where: • Method 1: Liquid transfer effect according to API 2000
C is a factor that depends on vapour pressure, average section [Link].2 + thermal inbreathing as per section
storage temperature, and latitude (see Table 2) [Link].3. In this approach, all tanks are assumed to be
Ri is the reduction factor for insulation empty, and the outflow of liquid occurs from all tanks when
Ri is 1 if no insulation is used there is no inflow to any of the tanks.
Vtk is the tank volume • Method 2: Liquid transfer effect based on API 2000 sec-
Vpe is the maximum rate of liquid discharge tion [Link].2 + thermal Inbreathing per section [Link].3.
Here, all tanks are assumed to be 50% empty, and the
In both the methods above, the calculation of tank outflow of liquid occurs only from one tank for a specific
thermal inbreathing requirement adopts a conservative service, with no inflow to that tank.
approach, assuming that tanks are empty and filled with air • Method 3: Inbreathing as per Annex F of API 2000. In this
before cooldown. However, it is believed that a more prag- method, all tanks are assumed to be 50% empty, and the
matic approach can be adopted, leading to a reduction in outflow of liquid occurs only from one tank for a particular
the normal nitrogen demand of the tank farm, which often type of liquid, with no inflow to that tank.
accounts for more than 20% of the overall refinery’s normal For detailed calculations, refer to Appendix 2, and a sum-
nitrogen demand. marised result of various methods is shown in Table 3.
In practice, storage tanks are typically not operated com- Table 3 shows that the nitrogen inbreathing requirement
pletely empty. They usually maintain some minimum inven- estimated using Method 1 is 1.7 times more than that esti-
tory levels. For product tanks in a specific service, one tank mated using Method 2 and 3.3 times more than that esti-
may be in receiving mode, another under certification, and mated using Method 3. Additionally, Method 2 provides an
another in despatch mode. Intermediate tanks are com- estimated nitrogen requirement 1.9 times higher than that
monly kept at around 50% level, while feed tanks are tar- estimated using Method 3. These comparisons highlight
geted to be kept full of inventory. the significant differences in nitrogen inbreathing estima-
It is pragmatic to consider a 50% level of inventory in the tions based on the different methods used.
tanks while estimating the nitrogen requirement for tank The thermal inbreathing requirements given by API 2000
inbreathing. This approach finds support in Annex F of API are approximately equivalent to a rate of change in ambient
2000, which states that: “If several tanks with a common temperature of 38°C per hour. While this may seem exces-
inert gas supply are divided so that no single tank has a sive, it reflects a change of about 10°C in 15 minutes, which
capacity exceeding 20% of the total capacity of all tanks, the is not uncommon as storm fronts move through. It also con-
calculated values may be reduced by 50%.” By applying this siders the impact of sudden cold rainfall on the shell of the
tanks.
Normal nitrogen inbreathing rate It is essential to consider that
the change in volume is com-
Method Liquid transfer (Nm³/h) Thermal inbreathing (Nm³/h) Total (Nm³/h) monly converted into equivalent
Method 1 31,082 283,551 314,633
volumetric rates based on air
Method 2 9,296 174,546 183,842
Method 3 9,296 87,273 96,569
at standard or normal condi-
tions. Consequently, the volu-
Table 3 metric rates may not appear as

4 February 2024 [Link]

equivalent displacement, especially when the assumed a) Flow reduction: As equipment is pressurised, a notable
operating or ambient temperatures do not align with stand- decrease in the nitrogen flow rate is observed, particularly
ard or normal conditions. The calculated inbreathing con- when vessel pressure exceeds half of the header pressures.
siders the assumption of ambient airflow through the tank This phenomenon can be seen in Figure 1, which highlights
vent, where it is customary to consider the ambient air to be the decline in flow rate under various pressures where the
at normal or standard conditions. nitrogen header pressure is at 9 bara.
In cases where a medium other than air is utilised for b) Excessive consumption: To estimate the nitrogen vol-
vacuum relief, it might be necessary to convert the rate to ume needed to achieve an oxygen content below 4 mol%,
an air-equivalent flow. Adjustments may be required for a comparative estimation is carried out for three different
inbreathing if the inbreathing medium significantly differs scenarios. In each scenario, the first cycle of nitrogen pres-
from air. However, no adjustments are necessary for nitro- surisation of the vessel, which is full of air, is considered.
gen, as the molecular weight of nitrogen (28.02) and air The vessel is pressurised from atmospheric pressure to
(28.96) exhibits only a marginal difference (3.3% variation different pressurisation scenarios. Here are three scenarios
in molecular weight). illustrating the differences:
In conclusion, facilities should assess the nitrogen
inbreathing requirement for tanks using the appropri- Scenario 1: Pressurised to 9 bara
ate method as described earlier. By doing so, they can Scenario 2: Pressurised to 5 bara
potentially achieve significant reductions in overall nitro- Scenario 3: Pressurised to 3 bara
gen requirement, leading to substantial savings in Capex
and Opex while ensuring safety and tank integrity are not Upon reaching the targeted pressure, if the oxygen con-
compromised. This prudent approach allows for efficient tent is below 4 mol%, the vessel is depressurised to its
resource allocation while maintaining the highest standards normal operating pressure of 3 bara, making it ready for
of operational safety. hydrocarbon introduction. Otherwise, it is brought down
to 1.5 bara and then pressurised to start the next cycle
Technical nuances and best practices in refinery units of pressurisation/depressurisation. This cycle continues
 Gaseous nitrogen for purging until the oxygen content meets the desired threshold, after
One of the most frequent uses of nitrogen is to purge equip- which the vessel is pressurised to 3 bara using nitrogen.
ment of explosive or hazardous vapours before lining up the The results are shown in Table 4.
vessel after maintenance or handing it over for maintenance. In Table 4, it can be observed that for Scenario 1, the
This is usually done using a pressure/de-pressure cycle. required nitrogen volume is the highest, being 1.5 times
The cycles needed to ensure the oxygen concentration is that of Scenario 2 and two times that of Scenario 3, even
below the Lower Explosive Limit (LEL) can be determined though only one cycle of pressurisation with nitrogen is
through the equation: required in Scenario 1.
Key takeaway: Limiting pressurisation to half of the nitro-
n = log[(Ci-Cn)/(Cf-Cn)]/log (Pf/Pi) Eq. 3 gen header pressure is both cost-effective and efficient, as
supported by the provided scenarios.
Where:
Cn = mol% oxygen in nitrogen  Start-up nitrogen for hydroprocessing units’
Ci = mol% oxygen initially in space to be purged high-pressure (HP) section
Cf = mol% oxygen finally in space to be purged The start-up of hydroprocessing units demands a sub-
Pi = Initial pressure in bara stantial amount of nitrogen, primarily for leak tests and
Pf = Final pressure in bara inertisation. For a typical 1,000 m3 volume hydroprocess-
n = Number of pressure/de-pressure cycles ing high-pressure loop, three cycles of pressurisations
and depressurisations between 1.5 barg and 5 barg are
For oxygen removal from equipment, a method involv- essential. The goal is to reach O2<0.5 mol%, necessitating
ing multiple pressurisation and depressurisation cycles is approximately 14,000 Nm3 of nitrogen.
employed by refiners. The equipment is first pressurised to
the maximum nitrogen header pressure. It is then depres-
surised from several points until a slight positive pressure
of approximately 0.5 barg is reached. This process of pres-
surising and depressurising is repeated until an oxygen
concentration below 4 mol% is achieved. For some cases,
like reformer or hydrotreater, target oxygen is less than 0.5
vol%. Once this level has been reached, the equipment is
pressurised to its normal operating pressure using nitro-
gen, ensuring it is prepared for hydrocarbon introduction.
The method of pressurising equipment to the maximum
nitrogen header pressure is associated with the following
drawbacks: Figure 1 Reduction of nitrogen flow rate vs vessel pressure

February 2024 5
[Link]
 Comparison table – pressurising to different levels

Scenario 1 Scenario 2 Scenario 3

1st Cycle 1st Cycle 2nd Cycle 1st Cycle 2nd Cycle
Cn = mol% oxygen in nitrogen 0.10 0.10 0.10 0.10 0.10
Ci = mol% oxygen initially in space to be purged 21.00 21.00 4.28 21.00 7.07
Cf = mol% oxygen finally in space to be purged 2.42 4.28 2.19 7.07 3.58
Pi = Initial pressure in bara 1.00 1.00 1.50 1.00 1.50
Pf = Final pressure in bara 9.00 5.00 3.001 3.00 3.0
n = Number of pressure/de-pressure cycles 1 1 1 1 1
Required nitrogen volume(Nm3) [vessel volume -V ] 8V 4 V 2V 2 V 2V
Total nitrogen volume for scenario (Nm³) 8 V 6V 4V
Note 1: Pressurisation is limited to 3 bara as oxygen content was below the target value of 4 vol%.

Table 4

Typically, hydroprocessing units undergo leak testing not practising vacuum pulling while inertising the HP loop.
at varying pressure levels using nitrogen. For the context This precaution helps to prevent potential hazards, such as
of this article, 20 barg is assumed as the maximum leak oxygen entering the system due to leaks. If oxygen does
test nitrogen pressure. Upon successful completion of this enter, it could react with hydrocarbons and catalysts in the
test, a nitrogen circulation at 20 barg is established. The system, as well as compromise the iron sulphide passiva-
process involves an initial cycle for detecting and rectify- tion layer.
ing leakages. Then, a second cycle is used to stabilise the Key takeaway: For optimal refinery operations, inertise
nitrogen circulation. Under conditions where the HP loop one hydroprocessing unit at a time, considering a minimum
is successfully tested at 5 barg and pressurised in the first peak demand rate of more than 3,000 Nm3/h for a 3in nitro-
cycle from 5 barg to 20 barg, the total nitrogen requirement gen line.
amounts to 35,000 Nm3, which includes leak test, inertising
and pressurisation.  Start-up nitrogen for hydroprocessing units
Frequently, hydroprocessing licensors specify a nitrogen low-pressure (LP) section
amount with a peak flow rate of 3,000 Nm3/h. The stand- As discussed in the previous section, hydroprocessing units
ard configuration for the nitrogen line to the HP loop is are identified as one of the major units demanding peak
3in, equipped with a full-bore globe valve. As per industry nitrogen during start-up. Because of this, the hydropro-
standards, the allowed depressurisation rate stands at 20 cessing unit’s LP section is commonly inertised with steam.
bar/hour, prompting operators to fully open the 3in globe Once sufficient steam out is completed, fuel gas is intro-
valve. With a 3in API 600 type globe valve with a Cv of duced after steam venting to the atmosphere is boxed up.
106 for a 1,500 lb rating, the flow results in 12,000 Nm3/h. The LP section is pressurised to the maximum fuel gas pres-
In contrast, when considering compressible flow through sure, which is typically around 4 barg. However, it should
a 3in 80 sch. pipe, the nitrogen flow rate is 8,000 Nm³/h be noted that if the normal pressure of the stripper exceeds,
where upstream nitrogen pressure is at 8 barg and down- say, 10 barg, the provision for nitrogen pressurisation to the
stream at atmospheric pressure. LP section is considered prudent since the nitrogen header
Key takeaway: The start-up nitrogen line for reactor loop pressure is approximately 8 barg. Such nitrogen pressurisa-
pressurisation should be sized to match licensor-prescribed tion is found essential during the start-up phase, especially
nitrogen demand in the utility summary. when the stripper requires pressurisation beyond 4 barg.
A noteworthy observation is that the peak nitrogen Key takeaway: It is recommended to provide a nitrogen
demand of a hydroprocessing unit significantly influences connection in addition to a fuel gas connection for the LP
the overall peak demand of the refinery, especially when the section.
refinery has multiple hydroprocessing units. At the design
stage, it is proposed by some refiners that the peak nitrogen  Gap control for blanketing gas for vessel
demand for hydroprocessing units be met using a tanker Normally, the surge drum level is permitted to fluctuate
connected directly at the unit level. By this method, the within a small band, typically within 5%, instead of main-
hydroprocessing units’ peak nitrogen demand is removed taining strict control. Due to this small fluctuation in level, a
from the overall refinery nitrogen balance. Consequently, swing in the drum pressure is caused. This pressure swing
potential savings are realised, as the requirement for a causes the blanketing valve to open when the level drops
nitrogen liquid tank and vapouriser is eliminated. However, and the vent to flare to open when the level rises. The con-
practical implementation can pose challenges due to lim- tinuous activation of these valves leads to a loss of nitrogen
ited access during unit shutdowns, especially during cat- from the surge drum. Typically, a dead band is incorporated
alyst loading/unloading. A prudent approach is to inertise into the surge drum pressure controller output, allowing
one hydroprocessing unit at a time. the pressure to fluctuate without opening the valves in the
To minimise risk, at present, hydroprocessing units are nitrogen and flare lines for minor demands.

6 February 2024 [Link]

Key takeaway: By implementing this gap control for all optimal purging protocols, sizing of nitrogen lines for HP
blanketing pressure controllers, the consumption of nitro- hydroprocessing units, actual peak nitrogen flow through
gen can be minimised. the utility station, necessity of nitrogen connection for the
LP section of hydroprocessing units, and effective control
 Flow through utility station of nitrogen mechanisms for gas blanketing.
Different approaches for determining the normal and peak In summary, this article endorses a customised approach
demand for refiners exist, and the determination of overall to managing nitrogen, considering actual operational
normal and peak demand is considered beyond the scope scenarios. It suggests a balanced use of both standard
of this article. Nonetheless, it is observed across refiners and alternative methods without compromising safety.
that for both normal and peak nitrogen demand, the nitro- Implementing these recommendations has the potential to
gen flow through at least one utility station is considered. result in significant cost reductions, more efficient resource
The nitrogen flow through a utility station is typically seen allocation, and the upholding of stringent safety norms.
to range from 200 Nm3/h to 400 Nm3/h. It was observed
that when considering compressible flow through a 0.75in
Appendices 1 and 2 can be viewed in the digital issue.
pipe, the nitrogen flow rate is 350 Nm³/h where upstream
nitrogen pressure is at 8 barg and downstream at atmos- Rajib Talukder is a Process Specialist in the Global Manufacturing
pheric pressure. Utility station sizes very often are 1in. Excellence department at Aramco, Saudi Arabia. He has more than
Key takeaway: The designer should consider the maximum 30 years of experience in process engineering and holds a [Link] in
possible nitrogen flow through a utility station while esti- chemical engineering from NIT Tiruchirappalli, India.
mating normal and peak nitrogen demand. Email: [Link]@[Link]

Conclusion Prabhas K Mandal is an Operations Engineer Specialist at Aramco.

This article emphasises key considerations for nitrogen He has more than 30 years of experience in petroleum refining, and
management in refinery operations at the design stage, supports front end design development for capital projects. He holds a
focusing on nitrogen blanketing in surge drums and tank- [Link] in chemical engineering and a [Link] in petroleum engineering.
age. The findings reveal that conventional guidelines, Email: [Link]@[Link]
notably those based on API 2000 standards, may lead to
overestimation of nitrogen requirements. Such overestima- LINKS
tion not only elevates nitrogen demand but also contributes
to operational inefficiency and waste. More articles from the following categories:
Beyond the scope of tank and surge drum blanketing, the Gasification
articles outlines crucial best practices for handling nitrogen LNG, NGL and GTL
throughout various units in a refinery. This encompasses

February 2024 7
[Link]