Pressure Relief System Reaction Forces

The Importance of Evaluating Existing Installations

Jason F. White P.E.
Smith & Burgess LLC
Jason.White@smithburgess.com
Bill Frenk
Frenk Water Solutions
Dustin Smith P.E.
Smith & Burgess LLC

Keywords: Pressure Relief, Overpressure, Reaction Forces, PSM Compliance, Action Item Resolution,
Relief Systems Design, Pipe Stress, Atmospheric Discharge, Mechanical Integrity
Abstract
Overpressure protection analysis has evolved significantly since the inception of the PSM standard, but
the mechanical stress applied to the piping during overpressure events appears to have been
overlooked. Recent events have led several facilities to request more attention be paid to the effects
stresses induced from relief device discharges have on their piping. The purpose of this study is to allow
an existing facility to focus resources on the relief device installations most likely to fail due to reaction
forces. A series of representative installations were evaluated in order to determine which parameters
associated with pressure relief have the strongest impact on the installations, with particular
concentration on the dynamic effects of the release. Screening criteria for identifying pressure relief
device installations which may exceed the allowable pipe stress levels were developed from these
systems. When evaluated against a petrochemical facilitys pressure relief systems and benchmarked
for its validity as a first pass tool to identify potential installations that may require further analysis or
additional physical supports, is was found that over 30% of the existing installations may require support
for reaction forces.
1. Introduction
Pressure relief devices control the amount and disposition of material during a process upset while
simultaneously protecting process equipment from damage due to overpressure caused by the upset.
The most common devices used for this purpose are pressure relief/pressure safety valves. Quite a bit
of engineering research, testing, and analysis has been performed to improve the ability to quantify the
suitability of a relief valve and the associated installation to protect equipment from overpressure. One
area that has less prescriptive requirements is analyzing the structural integrity of the relief device
installation during the emergency event. These installations are not designed for continuous flow; but

P: 713.802.2647 | F: 713.456.2181 | 7600 West Tidwell Road, Ste. 600 Houston, Texas 77040 | smithburgess.com

rather sporadic flow often times at choked/sonic conditions. This paper takes a brief look at the existing
evaluation of reaction forces for pressure relief valves, performs a detailed baseline analysis of typical
installations in order to develop a screening tool for evaluating an existing facility, and identifies the
results when tested against an existing petrochemical facility.
The purpose of this study is to limit the number of relief valves that require rigorous engineering
calculations to determine the adequacy of the installation. Often times it is assumed that pressure relief
valve installations are simple and easy to design; practical experience has shown that pressure relief
valve installations (particularly devices that discharge to the atmosphere) are the most easily
manipulated during the actual construction phase and often are not installed as they were intended.
See Figure 1 for two examples of installations that were likely not installed as designed (or designed at
all); clearly not all existing relief valve installations meet the specifications as recommended by industry
practice.

Figure 1 Examples of pressure relief valve installations where little or no engineering piping design was
performed or design was not followed.

P: 713.802.2647 | F: 713.456.2181 | 7600 West Tidwell Road, Ste. 600 Houston, Texas 77040 | smithburgess.com

2. Reaction Force Analysis Methodology

2.1

Reaction Forces General

During an overpressure event, the discharge of a pressure relief valve imposes a load, referred to as a
reaction force, on the collective installation. This force creates a bending moment that is both a
function of the quantity and state of the release and the physical layout of the piping installation: the
lever arm created by the installation. The stress caused by the reaction force is propagated into and
through the relief valve and then into the inlet piping and vessel nozzle, unless the system is properly
supported.
2.1.1

API Reaction Force Analysis

The American Petroleum Institute (API) provides guidance for determining pressure relief requirements
for pressure relief device installations. API 520 Part II (American Petroleum Institute, 2008) states that
pressure relief valve outlet piping should be independently supported and properly directionally aligned.
Stresses placed due to forced alignment of pressure relief valve piping are also mentioned; however,
that topic will not be elaborated in this paper. The authors' practical experience has demonstrated that
a significant portion of atmospheric relief devices do not have piping supports in place as described in
Figure 2.

Figure 2 - Recreation of Figure 7 from API 520 for a typical relief valve installation
Note 1 The support should be as close as possible to the centerline of the vent pipe.
Note 2 F = Reaction Force, A = Cross-sectional Area of discharge pipe.
API 520 Part II provides a calculation as a basis for the reaction forces in the event of a vapor or 2-phase
releases directly to the atmosphere. No discussion of the reaction forces developed during a liquid
release exists in this section. No guidance with respect to applying these results or determining if an
installation is acceptable are presented; the burden is placed on the designer to ensure that the
installation is appropriate. While this may be reasonable for the design of new facilities, evaluating the
adequacy of existing facilities becomes much more complicated.

P: 713.802.2647 | F: 713.456.2181 | 7600 West Tidwell Road, Ste. 600 Houston, Texas 77040 | smithburgess.com

The formulas as from API RP 520 Part 2 are listed below for relief devices discharging to the atmosphere:

W
kT
AP
366 (k 1) M

API 520 pt II 4.4.1.1 U.S. customary units for vapor

relief reaction forces

x (1 x )
W2

A( Pe Pa )
6
2.898E10 A g
l

API 520 pt II 4.4.1.2 U.S. customary units for 2phase relief reaction forces

Where
F = Reaction force at the point of discharge
to the atmosphere, lbf
W = Flow of any gas or vapor, lbm/hr
Cv = Specific heat at constant volume
M = Molecular weight of the process fluid
P = Static pressure within the outlet at the
point of discharge, psig
rg = Vapor density at exit conditions, lbm/ft3
Pe = Absolute pressure at pipe exit, psia
2.1.2

k = Ratio of specific heats (CP/CV) at the outlet

conditions
Cp = Specific heat at constant pressure
T = Temperature at the outlet, R
A = Area of the outlet at the point of
discharge, in2
x = Weight fraction vapor at exit conditions
rl = Liquid density at exit conditions, lbm/ft3
Pa = Absolute ambient pressure, psig

DIERS Reaction Force Analysis

The Design Institute for Emergency Relief Systems (DIERS) provides much of the same guidance for the
considerations for reaction forces and the determination as to the acceptability of a relief device
installation. (Fisher, 1992) Some additional recommendations are provided with regard to the
suggested piping layout to avoid excessive lever arms as recreated in Figure 3.
Additionally, emphasis is placed on the importance of evaluating the reaction forces for all credible
overpressure contingencies, not simply the controlling contingency. This is very important because the
physical properties are not always the same, and in some cases, the controlling contingency for sizing
may be a vapor stream while the controlling case for the reaction forces may be an all 2-Phase stream.

Figure 3 Example of how even the smallest modification in piping design can have a significant effect
on the resulting reaction forces. In this case, the system on the left has significantly more stress due to
the increased lever arm and direction of discharge when compared to the system on the right.

P: 713.802.2647 | F: 713.456.2181 | 7600 West Tidwell Road, Ste. 600 Houston, Texas 77040 | smithburgess.com

3. Reaction Force Case Study Analysis

A piping system may respond far differently to a dynamic load than it would to a static load of the same
magnitude. Static loads are those which are applied slowly enough that the piping system has time to
react and internally distribute the loads, thereby remaining in equilibrium.
With a dynamic load a load which changes quickly with time the piping system may not have time to
internally distribute the loads, so forces and moments are not always resolved, resulting in unbalanced
and potentially concentrated loads and pipe movement.
The typical action of relief valve venting is an impulse load where the flow rate and associated forces
ramps from nominally zero to some value, remains relatively constant for the duration of the release,
and then ramps down to zero again. When the relief valve opens, the discharge fluid creates a jet force
which acts on the piping system; this force increases to its full value, from zero, over a time frame
similar to the opening time of the valve. The relief valve remains open until sufficient fluid is vented to
relieve the over-pressure situation. When the valve then closes, the reduction in flow corresponds to
the loss of the jet force over the closing time of the valve.
Multiple relief valve piping configurations were analyzed for both static and dynamic conditions using
CEASAR II. The stresses calculated during the analysis were checked against the allowable stresses per
ASME B31.3 - 302.3.5. Additionally, the analysis was used to determine if a flange leak was likely. In all
cases, the dynamic condition was determined to be the governing condition for the structural integrity
of the piping system.
The leakage check looks at the tendency of the flanges to separate under the applied piping loads. ASME
B31.3 does not address flange leakage directly. The purpose of this analysis was to determine piping
failure and not flange leaks. From extensive tests, it has been determined that even under unusually
severe bending stresses, flange assemblies did not fail in the flange proper, by fracture of the bolts, or
by leakage across the joint face. Structural failure occurred almost invariably in the pipe adjacent to the
flange, and in rare instances, across an unusually weak attachment weld. Leakage well in advance of
failure was observed only in the case of threaded flanges. (Peng, 2009). This would suggest that the
structural integrity of the piping is the major area of concern for any stress analysis and should be
considered over flange leakage. This is the basis for these evaluations.
A variety of process connections have been observed in field installations: welding reducing tees,
weld/thread O-Lets, and unreinforced Stub-In connections. Unreinforced Stub-In connections result
in the highest ratio of calculated to Code allowed stresses, followed by O-Lets and then welding
reducing tees for the same relief system applied piping loads. The model used in the current evaluations
has been confined to welding reducing tees.
The allowable Code stresses are below the yield and well below the tensile as indicated below for
commonly used carbon steel materials. The net result is that there is approximately 19% to 24% safety
factor between the Code allowable for Occasional Loading and the yield point where the material
begins to fail. The relief valve models were evaluated to establish relief pressures at which the
calculated stresses were within 5% of each of the Allowable Occasional, Yield, and Tensile stress.

P: 713.802.2647 | F: 713.456.2181 | 7600 West Tidwell Road, Ste. 600 Houston, Texas 77040 | smithburgess.com

Table 1 Sample of allowable stresses used in screening study

Material

A 234 (tee)
API 5L B (Pipe)
A105 (Flange)
3.1
3.1.1

Allowable Stress
B31.3 Table A-1
(psi)
23,300
20,000
21,900

Allowable Stress
Occasional Load
(psi)
30,990
26,600
29,130

Yield Stress
B31.3 Table A-1
(psi)
40,000
35,000
36,000

Tensile Stress
B31.3 Table A-1
(psi)
70,000
60,000
70,000

Software Used and modeling details

Caesar II 5.30

The relief valves were modeled as an Open Discharge; with a vertical pipe discharging directly to
atmosphere. The process connection is mounted on a pipe header with a welding reducing tee. This
arrangement was chosen to provide a more realistic representation of typical installations together
with the inherent flexibility of the tee / header connection. The vent pipe is the same diameter as the
outlet connection on the valve and is unsupported at the elbow with a 6-0 long vertical vent pipe.

Figure 4 - Sample of the model basis as developed in Caesar II.

3.1.2

SolidWorks Flow Simulation

SolidWorks was used to determine the physical properties along the vent pipe required to calculate the
thrust and momentum forces.
Average velocity along the vent pipe
Average temperature across the outlet of the vent pipe
Average velocity at the elbow
The relief valve was modeled as an orifice at the end of a converging nozzle. The orifice was set to
produce the capacity calculated by the Caesar Relief Valve Symmetry analysis for the given inlet
conditions using the certified orifice size.

P: 713.802.2647 | F: 713.456.2181 | 7600 West Tidwell Road, Ste. 600 Houston, Texas 77040 | smithburgess.com

Assumptions: The following assumptions were made regarding the analysis.

Process fluid is vapor.
Manufacturers certified orifice diameter from National Board Relief Device Certification NB -18
were used in place of standard API orifice diameters to provide more realistic discharge flow.
Crosby JOS valve orifice data was used.
Valve opening and closing time is 8.0 milliseconds. Venting will last for (1) one second. While
these numbers are specific to the valve manufacturer, they appear to be typical throughout the
relief valve industry.
Wind loadings were not considered.
All piping considered to be Schedule 40 carbon steel.
Relief valve inlet flanges: as required for process considerations.
Relief valve outlet flanges: ANSI RF 150#.

Figure 5 Sample of velocity profile output from SolidWorks Flow Simulation

4. Screening Study and Results
The objective of developing the screening tool was to provide a fairly quick method to identify relief
valves in a facility that were likely to need either support or more detailed analysis to verify the
adequacy of the existing installation. For the purposes of simplification, the following assumptions were
made:

P: 713.802.2647 | F: 713.456.2181 | 7600 West Tidwell Road, Ste. 600 Houston, Texas 77040 | smithburgess.com

All relief valves discharging to a closed disposal system are adequately supported for an individual
release Relief valve installations that discharge into a closed system are by definition supported by at
least the point of discharge, and the purpose of this screening is to identify relief valves that require
support, not to evaluate the adequacy of existing supports. Because of the complexity of a supported
common disposal system, they are excluded from the scope of this study.
All liquid and 2-phase relief contingencies require detailed analysis It appears water hammer is a
much larger concern for liquid and 2-phase releases than reaction forces, and is therefore an item to be
evaluated outside of reaction forces. Additionally, the dynamic effects of flashing flow create far too
many variables to include in a simplistic screening.

API Orifice Letter

All non-standard pressure relief valve sizes require detailed analysis While this statement may not be
true for every installation, for the purposes of developing an automated tool to identify relief valves that
may need detailed engineering, the obvious decision was to flag any installation that falls out of the
normal range. For the purposes of this study, standard us defined by the flanged relief valve sizes as
listed in API 526 and shown in Figure 6.

D
E
F
G
H
J
K
L
M
N
P
Q
R
T

1X2

1.5X2

Inlet Diameter (in.) X Outlet Diameter (in.)

1.5X2.5 1.5X3 2X3 2.5X4 3X4 4X6

6X8

6X10

8X10

Figure 6 Relief valve size configurations evaluated per API 526

Pressure relief valves installations can be characterized as either Typical or Complex Because the
generic screening methodology was performed using a typical pressure relief valve configuration (as
seen in Figure 4), some method of identifying configurations with more complex piping had to be
identified. For the purposes of this study, any piping configuration containing more than charge of
direction fitting (elbow, 45 bend, branch tee, etc.) were considered to be complex, using Figure 3 as
the basis for this assumption.

P: 713.802.2647 | F: 713.456.2181 | 7600 West Tidwell Road, Ste. 600 Houston, Texas 77040 | smithburgess.com

Pressure relief valves installed and sized for the external fire contingency only will not require
reaction force evaluation While this may seem counter intuitive, as external fire is the prevailing
overpressure contingency in pressure relief systems design, it is proposed here that a pressure relief
valve installation cannot be deemed adequate by a reaction force analysis in the event of an external
fire. The affect heating from a fire would have on the relief device are unknown and can be more
significant on the relief device installation (particularly the outlet piping and the valve body itself) than
the stresses caused by the flow rate. It is proposed that a relief valve installation failing due to reaction
forces is more likely controlled by the reduction in tensile strength of the installation due to the heat
input than the system design.
Pressure relief valves installed and sized for the liquid hydraulic expansion contingency only will not
require reaction force evaluation Liquid hydraulic expansion cases are often nominal rates for which a
small thermal pressure relief valve is installed, and in many cases possesses a capacity far greater than
the required relief load. Additionally, the non-steady-state nature of a thermal expansion event tends
to results in the unsustained releases that do not develop typical fluid flow characteristics. Taking these
factors into account, it was determined that hydraulic expansion scenarios do not require pipe stress
screening.
4.1 Qualitative Screening
The screening study was divided into two phases; one being qualitative screening against assumptions
as set forth above, and the other being against the criteria set for in the base case study for each relief
valve size. The qualitative step was performed stepwise as a discussion tree as represented in Table 2.
Table 2 Stepwise results of decision tree of qualitative results to determine relief devices that require
detailed engineering analysis.
Qualitative Step
Starting Point
External Fire Only
Thermal Expansion Only
Discharge to Closed System
Non Standard Device Sizes
Liquid or 2-phase Relief

Relief Valves
Remaining
189
186
168
157
152
112

Relief Valves
Requiring Analysis
0
0
0
0
5
45

Relief Valves Not

Requiring Analysis
0
3
21
32
32
32

Therefore, the qualitative screening step identified 32 installations that are acceptable as is, 45 that
require more detailed analysis (including the potential review of why two-phase/liquid releases are
being sent to the atmosphere), and 112 installations that are not covered by this screening..

P: 713.802.2647 | F: 713.456.2181 | 7600 West Tidwell Road, Ste. 600 Houston, Texas 77040 | smithburgess.com

4.2 Quantitative Screening

The second stage of the screening was developed based on generic pressure relief device installations.
The intention was to draw a line separating pressure relief valve installations into three categories:
installations predicted to be acceptable as installed, installations that would require detailed
engineering analysis to determine the adequacy of the installation, and installation that are expected to
require proper piping support if a detailed engineering analysis was performed; the results of this
screening are found in Table 3. This screening was performed against all 189 relief device installations
rather than the remaining 112 from the qualitative screening, due to the fact that three possible results
exist rather than the two previously used. A relief valve installation that was flagged as needed a
detailed analysis in the qualitative step may be identified as a device that is predicted to require support
irrespective of a detailed analysis. Additionally, a relief device sized only for fire may be identified as an
installation requiring support even before the effects of temperature change can be accounted for.
In order to perform this screening the existing pressure relief valve installations were divided into
typical and complex groups as described previously. For typical installations a threshold value of
90% was used when comparing the installation to screening tool generated stresses for each relief valve
size; meaning that relief valves having a relief pressure within 90% of the threshold for occasional
loading were flagged as to require detailed engineering analysis and relief pressures exceeding 90% of
the threshold for yield stress were flagged as expected to require support regardless of the detailed
analysis. For complex pressure relief valve installations the threshold value was lowered to 70% of the
occasional loading/yield stress limits respectfully.
Table 3 Quantitative Screening Results for Pressure Relief Valve Installations based on Complexity of
the Installation
Installation Type
Typical
Complex
Total

# of installations
145
58
189

Requiring detailed analysis

4
5
9

Require Support
15
13
28

The results of the quantitative screening indicate that most of the valves predicted to possible require
support (28 of the 37 identified) fall into the category of those predicted to exceed the yield stress and
therefore identified as installation that do not require detailed analysis to determine if support is
required.

P: 713.802.2647 | F: 713.456.2181 | 7600 West Tidwell Road, Ste. 600 Houston, Texas 77040 | smithburgess.com

4.3 Overall Results

The results of both screening studies were combined to create an actionable list of items for ensuring
the physical integrity of the pressure relief valve installations, which can be seen in Table 4. For the
facility studied, the aggregate of the two screenings predicted that 28 pressure relief installations would
require support (even if a detailed engineering study were performed), and an additional 34 pressure
relief valve installations would require a more detailed engineering study to determine the adequacy of
the installations. Overall, more than 30% of the relief valves studied were found to require action.
Table 4 Overall Results based on Reaction Force Screening
Action Item
Relief Devices Requiring Support
Relief Devices Requiring Engineering Analysis
Installations Predicted to be Adequate with Respect to Reaction Forces
Total

Quantity
28
34
127
189

A sample from each of the three categories was taken and detailed analysis was performed to verify
these results. Of that sample all relief device installations predicted to require support did indeed
require support to avoid exceeding the yield stress; likewise all sampled installations predicted to be
adequate were found to be adequate.
Of the sampled devices predicted to require detailed
engineering analysis, all but one was resulted in exceeding the yield stress, and that installation did
exceed the allowable stress. The purpose of this study was to provide a solid screening tool in order to
prevent the cost of performing detailed engineering evaluation on every relief device installation, and
the end result proved to succeed at this. In some cases, it may be more cost effective to simply support
pressure relief valve installations for which a detailed study is suggested rather than perform the
detailed study.
5. Conclusions
Overpressure protection analysis has evolved significantly since the inception of the PSM standard, but
the mechanical stress applied to the piping during overpressure events appears to have been for the
most part overlooked. Criteria for identifying pressure relief device installations which may exceed the
allowable stress levels were developed from these systems. This criteria was then evaluated against a
petrochemical facilitys pressure relief systems and benchmarked for its validity as a first pass tool to
identify potential installations that may require physical supports. For the facility studied, ~2/3 of the
pressure relief valve installations were predicted to be adequate with respect to reaction forces with the
remaining installations being broken into two categories, those requiring support, and those requiring
further analysis. Proving that in practice, a significant percentage of pressure relief valve installations do
not meet the desired structural integrity when considering reaction forces. This study demonstrates a
screening tool that allows plants to focus resources on the relief valve installations most likely to fail due
to reaction forces.

P: 713.802.2647 | F: 713.456.2181 | 7600 West Tidwell Road, Ste. 600 Houston, Texas 77040 | smithburgess.com

6. References
I.
II.
III.
IV.
V.
VI.
VII.

Relief System Reaction Forces in Gas and Two-phase Flow, 1991, 25th Annual AIChE Loss
Prevention Symposium
Emergency Relief System Design Using DIERS Technology, 1992, New York, NY, American
Institute of Chemical Engineers
Thrust Force Calculations for Pressure Safety Valves, 2006, Process Safety Progress, 203-213
Pipe Stress Engineering, 2009, New York, NY, American Society of Mechanical Engineers
American Petroleum Institute, API Standard 520, Sizing, Selection, and Installation of Pressurerelieving Devices in Refineries, API
American Society of Mechanical Engineers, 2004, B31.3 Process Piping Guide, American Society
of Mechanical Engineers
CCPS - Center for Chemical Process Safety, 1998, Pressure Relief and Effluent Handling Systems,
New York, NY, American Institute of Chemical Engineers

P: 713.802.2647 | F: 713.456.2181 | 7600 West Tidwell Road, Ste. 600 Houston, Texas 77040 | smithburgess.com