You are on page 1of 10

Proceedings of the ASME 2022

14th International Pipeline Conference


IPC 2022
September 26-30, 2022, Calgary, Alberta

IPC2022-87102

A RISK-BASED DESIGN APPROACH FOR UNCASED PIPE UNDER ROADS AND


RAILWAYS

Hafeez Nathoo, Maher Nessim, Mark Stephens

C-FER Technologies, Edmonton, Alberta

ABSTRACT RR risk of failure


A risk-based pressure design approach has been developed TI interruption time (days)
for uncased pipe under roads and railways. Similar to the V traffic volume (vesicles/day)
approach currently used in Canadian Standard Association’s Vt freight traffic volume (trucks/lane/day or trains/day)
Standard Z662, the approach uses a set of hoop stress factors to VSL value of a statistical life ($)
calculate the minimum wall thickness from the pipe’s pressure, W assessment area width (m)
diameter, and specified minimum yield strength. The hoop stress XI interruption cost ($)
factors were calibrated to meet specified reliability targets a adjusted population density (people/ha)
considering the risk factors specific to roads and railways, which r population density (people/ha)
include elevated probabilities of mechanical damage due to xI delay cost ($)
higher construction activity rates, safety impact on road users,  incremental average population density over the
and potential costs of traffic interruption in case of a pipeline assessment area (people/ha)
failure. The hoop stress factors are defined as a function of the
safety class, which is determined according to the approach 1. INTRODUCTION
described in a companion IPC paper. The Technical Committee (TC) for the Canadian Standards
This paper describes the development approach and Association’s (CSA’s) Standard Z662 on the design and
provides a comparison between its results and the designs operation of pipeline systems (the “Standard”) formed a class
obtained from the current CSA Z662 approach. An analysis location task force (CLTF) to review and propose improvements
confirming adequacy of the resulting wall thicknesses to to the class location system and location factors used in Clause 4
withstand normal traffic loads is also presented. The approach of the Standard. The CLTF determined that a new risk-based
is being proposed as an alternative to the hoop stress factors pressure design approach is required. The risk-based design
currently used in CSA Z662. approach was developed under a joint industry project (JIP). The
Keywords: risk-based design, pressure design, crossings, goal of the JIP was to develop a risk-consistent safety class
roads and railways system and use it as a basis for defining a set of pipeline design
hoop stress factors that achieve known, acceptable, and
NOMENCLATURE consistent public safety and environmental protection levels for
pf probability of failure due to equipment impact for all pipelines and service fluids within the scope of Clause 4 of
‘general’ pipe the Standard. The hoop stress factor is defined as the ratio
rh hit rate adjustment factor between the hoop stress under the design pressure and the
s average traffic speed (km/h) specified minimum yield strength (SMYS).
wl lane width (m) As part of this work, an approach was developed to select
A assessment area (hectare (ha)) the wall thickness for pipe under roads and railways considering
Ar road surface within assessment area (ha) the relevant risk factors at such locations, including elevated
C consequence indicator probabilities of mechanical damage due to higher construction
CI adjusted consequence indicator activity rates, safety impact on road users, and potential costs of
L assessment area length (m) traffic interruption in case of a pipeline failure.
Lr length of road within assessment area (m) This paper describes a pressure design approach developed
N number of fatalities for uncased pipe under roads and railways and provides
Nl number of lanes comparisons between the wall thicknesses resulting from the
Np average number of occupants per vehicle

1 © 2022 by ASME
risk-based approach and the approach currently used in the class. The adjusted population density, a, can be calculated as
Standard. follows (see Figure 1):
×
𝜌 = 𝜌 + ∆𝜌 = 𝜌 + (1)
2. APPROACH AND SCOPE
The risk-based pressure design approach involves two steps: where  is the incremental average population density over the
1. Define a safety class based on the severity of pipeline failure assessment area (people/ha),  is the population density not
consequences [1]. The safety class is a function of the considering the road (people/ha), A is the assessment area (ha),
pipeline attributes, such as diameter and pressure; service r is the density of road users (people/ha), and Ar is the road
fluid properties, such as composition; and location surface area within the assessment area (ha).
characteristics, such as population density and The road surface area within the assessment area can be
environmental sensitivity. To achieve uniform risk, an calculated from
allowable failure probability that is inversely proportional to
× ×
the severity of the failure consequences, is assigned to each 𝐴 = (2)
,
safety class. The safety class replaces the class location
system currently used in the Standard. where Nl is the number of lanes, wl is the lane width (3 m), and
2. Define the hoop stress factor for pressure design as a Lr is the length of road within the assessment area.
function of the safety class and pipeline attributes, such as The population density adjustment is based on the road
diameter, pressure, SMYS, fracture toughness, and length, Lr, within a 1,600 m by W assessment area [3] centred on
equipment impact hit rate. The hoop stress factor for each the point of intersection between the pipe centreline and the road
safety class was calibrated to meet the applicable allowable centreline. Lr equals the assessment area width, W, if the road is
failure probability [2]. perpendicular to the pipeline, and equals the assessment area
length, L (1,600 m), if the road is parallel to the pipeline. For
To adapt this approach to uncased pipeline segments under
roads oriented at other angles, Lr can be calculated from simple
roads and railways, the additional risk factors specific to these
geometric relationships.
pipeline segments were identified and incorporated in the
definition of the safety class in step 1. The elevated safety class Wr
can then be used to select the appropriate hoop stress from the
relationships in step 2.
The road- and rail-specific risk factors considered include
the following:
• Additional failure consequences, including the safety Assessment area, A
impact on road users and the societal costs of traffic W LR
Pipeline
disruption.
• Increased frequency of equipment impact due to
Wr
excavations related to maintenance of the road or railway, as
well as other utilities in the same corridor.
Road
While the Standard has specific requirements to address the
traffic loads for railways, similar requirements are not included a) Pipe Perpendicular to Road (Crossing)
in the Standards for roads. Designers appear to rely on the L
pressure design hoop stress factors in the Standard to ensure that
the pipe wall thickness is adequate to withstand normal road Assessment area, A
traffic loads. To ensure that the risk-based design approach does W
Pipeline
not result in a gap in this area, the risk-based designs for pipe
under roads were checked to ensure their adequacy under normal Wr
traffic loads.
Road LR
3. SAFETY OF ROAD USERS
b) Pipe Parallel to Road
3.1 Approach
Since railways (excluding light rail, which is not currently FIGURE 1: ILLUSTRATION OF ROAD SURFACE
considered in the Standard) are used primarily for freight, the WITHIN THE ASSESSMENT AREA FOR ROADS
probability that a pipeline failure will occur simultaneously with
the presence of a passenger train is assumed to be negligible, and The population density, r, in people/ha for the road can be
safety of users is considered for roads, but not for railways. calculated from
Safety impact is accounted for by including road users in × × ,
calculating the population density used to determine the safety 𝜌 = (3)
× × × ×

2 © 2022 by ASME
where Np is the average number of occupants per vehicle, V is as a basis for defining the safety class [1]. C is related to the
the traffic volume (vehicles/day), and s is the average traffic expected number of fatalities, N, by
speed (km/h). Equation (3) calculates the average distance
between vehicles along the road as the ratio between the average 𝐶= (4)
. ×
speed and traffic volume. It then calculates the density as the An adjusted consequence indicator, CI, that includes the
average number of vehicle occupants divided by the road surface effect of traffic interruption can be calculated from
area occupied by each vehicle.
𝐶 =𝐶+ (5)
. × ×
3.2 Road Population Density where XI is the interruption cost in $/incident and VSL is the
The road population density depends on site-specific value of a statistical life. VSL is estimated to be CAD8.63 million
information, including the number of lanes, traffic volume, and based on the mean value provided by the Treasury Board of
average road speed. Since this information is not currently Canada [9], adjusted for inflation to 2021 dollars. Based on the
collected by designers, conservative default values were defined same rationale discussed for population density, a single value of
for two road categories that can be used directly in the design CI is required to define the safety class for a crossing, whereas
approach. Alternatively, a designer can collect the required for pipe parallel to the road or railway, CI must be defined as a
information and use it to develop a more accurate design basis. function of station and used as a basis for segmenting the
The default road population densities were calculated as pipeline by safety class.
simple averages of the individual densities for a sample of roads
in different Canadian jurisdictions [4-7] (see Tables 1 and 2), TABLE 1: TRAFFIC DATA USED IN DEFINING THE
assuming an occupancy rate of 1.7 people/vehicle [8]. The POPULATION DENSITY FOR MAJOR ROADS
resulting population densities were rounded up to 35 people/ha
Traffic Pop.
for major roads, consisting of arterial roads, freeways and Avg.
Road volume No. of Density
Source Speed
expressways; and 20 people/ha for minor roads, consisting of Category (vehicles/ Lanes (people/
(km/h)
local and collector roads. day) ha)
North
Arterial 50,000 50 2 118.1
3.3 Application Vancouver
Determination of the safety class (as described in Nessim et City of
Arterial 20,000 40 2 59.0
al. [1]) depends on the configuration of the pipe in relation to the Toronto
road as follows: North
Arterial 20,000 50 2 47.2
 Crossings (perpendicular or angled). A single value of a, Vancouver
which equals the sum of  and the maximum value of  City of
Arterial 20,000 70 2 33.7
along the length of the crossing, is used to define the safety Hamilton
class for the crossing. Alberta
Arterial 30,000 60 4 29.5
Transport
 Pipe parallel to the road. The population density, a, should
be calculated along the pipeline and used, instead of , to Alberta
Arterial 15,000 60 2 29.5
Transport
define the safety class as a function of station using the
approach described in Nessim et al. [1]. This calculation City of
Arterial 20,000 50 4 23.6
Toronto
assumes that the entire road surface area is within the
assessment area and can be adjusted to consider the actual Alberta
Arterial 30,000 110 4 16.1
Transport
area if the road is only partially within assessment area.
Alberta
Arterial 10,000 100 2 11.8
Figure 2 shows the population density adjustment factor, Transport
defined as the ratio between the population densities calculated City of
Arterial 10,000 80 4 7.4
with and without accounting for road travellers for natural gas Hamilton
pipelines. The relative adjustment increases as a function of the City of Freeway/
5,000 80 2 7.4
number of lanes and decreases as a function of the surrounding Hamilton expressway
population density. The adjustment is more significant if the pipe Alberta Freeway/
60,000 80 4 44.3
is parallel to the road because a larger area of the road surface is Transport expressway
included in the assessment area, and is negligible for City of Freeway/
40,000 80 4 29.5
perpendicular crossings if the surrounding population density is Toronto expressway
greater than 1 person/ha. Alberta Freeway/
10,000 90 4 6.6
Transport expressway
4. IMPACT OF TRAFFIC INTERRUPTION Alberta
Arterial 10,000 110 4 5.4
4.1 Approach Transport
Interruption cost is incorporated in the consequence Average 31.3
parameter (referred to as the consequence indicator, C) and used

3 © 2022 by ASME
TABLE 2: TRAFFIC DATA USED IN DEFINING THE Equation (6) assumes that, during a road closure, trucks will find
POPULATION DENSITY FOR MINOR ROADS alternate routes or be used for alternate deliveries and, therefore,
Traffic Pop. the delay cost per truck affected is independent of the duration
Avg.
Source
Road volume
Speed
No. of Density of the closure.
Category (vehicles/ Lanes (people/ Truck delay cost, xI, was determined from a report by the
(km/h)
day) ha)
National Cooperative Highway Research Program [13]. The
City of report calculates direct interruption cost from operational costs
Collector 8,000 40 1 47.2
Toronto
(including fuel and driver pay) and revenue losses to the trucking
North companies and freight-dependent operations, based on detailed
Collector 8,000 30 2 31.5
Vancouver
information regarding the local transportation network, freight
Alberta operating costs, alternate routes, and potential long-term
Collector 10,000 50 2 23.6
Transport
impacts. Two case studies from Washington State were selected
City of to estimate a representative value of xI: 1) a closure of Interstate 5
Collector 12,000 60 2 23.6
Hamilton
due to an extreme rain event, and 2) a closure of Interstate 9 due
City of to heavy snowfall. I-5 and I-9 have daily traffic volumes of 8,333
Collector 5,000 60 2 9.8
Hamilton
and 7,134 vehicles/lane, respectively [7], and were considered
City of representative of Canadian major roads, which have an average
Collector 5,000 80 2 7.4
Hamilton
traffic volume of 8,333 vehicles/lane/day based on the data in
Alberta Table 1. The average unit delay cost calculated from the two
Collector 1,000 90 2 1.3
Transport
incidents (see Table 3) is xI = $615 per truck, in 2008 USD.
North
Local 20,000 30 2 78.7
Vancouver
40
City of
Local 2,500 40 1 14.8 5 Lanes
Toronto 35
4 Lanes
Population Density Adjustment Factor

Alberta 3 Lanes
Local 1,000 40 1 5.9 30
Transport 2 Lanes
25
North One
Local 1,500 30 2 5.9
Vancouver
20
City of
Local 3,000 60 2 5.9
Hamilton 15

City of 10
Local 1,000 50 2 2.4
Hamilton
5
City of
Local 1,000 60 2 2.0
Hamilton 0
0.01 0.1 1 10 100
Alberta
Local 1,000 70 2 1.7 Surrounding Population Density (People/Ha)
Transport
Average 17.4 a) Pipe Perpendicular to Major Road
600
NPS 4, at 6 MPa - 4 lanes
4.2 Interruption Costs NPS 4, at 6 MPa - 2 lanes
Population Density Adjustment Factor

Quantifying the societal impact of road and railway 500


NPS 24, at 16 MPa - 4 lanes
interruptions is a complex situation-specific problem as NPS 24, at 16 MPa - 2 lanes
indicated by the relevant literature [10-12]. To develop a simple 400 One

approach suitable for design, it is assumed that the direct


300
economic impact associated with freight traffic interruption is a
reasonable indicator of the overall societal impact of both road
200
and railways closures; other societal impacts related to private
traveller delays are not considered. Generic order-of-magnitude
100
interruption cost estimates from the literature were used, with the
recognition that pipeline-specific values can be developed and
0
used by designers where a more refined approach is needed. 0.01 0.1 1 10 100
For roads, the freight delay cost per lane of traffic, XI, is Surrounding Population Density (People/Ha)

estimated as follows: b) Pipe Parallel to Minor Road


𝑋 =𝑥 ×𝑉 ×𝑇 (6) FIGURE 2: POPULATION DENSITY ADJUSTMENT
where xI is the delay cost per truck ($/truck), Vt is the traffic FOR ROADS VERSUS SURROUNDING POPULATION
volume (trucks/day/lane), and TI is the interruption time (days). DENSITY

4 © 2022 by ASME
TABLE 3: ROAD UNIT DELAY COSTS DETERMINED TABLE 4: UNIT DELAY COSTS BY TRAIN TYPE FROM
FROM THE I-5 AND I-9 CLOSURES LOVETT ET AL. [15]
Daily Direct Weighted Unit
Unit Delay Unit Delay Cost
Truck Duration Total No. Delay Train Type Weight Delay Cost, xI
Hwy Cost, xI (USD/train-day)
Traffic of Delay of Trucks Costs (USD/train-day)
Closed (USD/truc
(trucks (days) Affected (Millions
k) Unit Train 57,600 31% 17,856
/day) of USD)
I-5 10,000 4.0 40,000 24.87 622 Manifest train 60,000 31% 18,600

I-9 6,500 3.7 24,104 14.75 611 Intermodal 86,400 38% 32,832
Average Unit Delay Cost (USD/train-hour) 69,288
Freight traffic volume per lane, Vt, was calculated for major
and minor roads as the average of the traffic data in Tables 1 and The average rail traffic, Vt, at a randomly selected location
2, assuming that freight traffic volume is 15% of the overall daily along a railway track was determined from the 2019 Canadian
vehicle volume [7]. Rail Trends report [16], which states that trains travelled
Average road closure time, TI, was estimated from 17 110,354,300 km in 2018 on a total track length of 41,682 km.
pipeline failures near roads that were reported in public sources Assuming that travel is uniformly distributed throughout the year
in the last six decades [14], which shows that the average and along the total track length, the expected train traffic volume
interruption time is approximately 30 hours (1.25 days). on a randomly selected point of track is 7.24 trains per day.
The values of xI, Vt, and TI were used in Equation (6) to In the absence of data on interruption times for railways due
calculate a per incident cost of approximately USD848,000 per to pipeline failures, TI was assumed to be the same as for roads
lane for major roads and USD353,000 per lane for minor roads. (i.e. 1.25 days per incident).
For railways, the delay cost, XI, is estimated as follows: The values of xI, Vt, and TI were used in Equation (7) to
𝑋 = 𝑥 × (𝑉 × 𝑇 ) × 0.5 𝑇 = 𝑥 × 𝑉 × 0.5 𝑇 [7] calculate a delay cost of approximately USD393,000 per
incident.
where xI is the delay cost per train per unit time ($/train-day), Vt
is the traffic volume (trains/day), and TI is the interruption time
4.3 Application
(days). To reflect the linear nature of railways, Equation (7)
Interruption costs were converted to 2021 CAD based on the
assumes that all train traffic will be delayed at the failure location
USD exchange rate and consumer price index (Table 5) and used
until the railway is restored. In this equation, (𝑉 × 𝑇 ) equals the
in Equation (5) to calculate CI. Figure 3 shows the relative
number of trains affected. The delay time depends on the location magnitude of the consequence indicator with and without
of the train at the time of closure, with the train arriving at the interruption cost (CI/C) as a function of population density for a
failure site immediately after closure, being delayed by the full two-lane minor road, a four-lane major road, and a railway. The
closure time, and the train arriving immediately before re- low vapour pressure (LVP) pipeline in the figure is assumed to
opening, not being delayed. Assuming that trains are equally be in a non-DGA (where DGA is a designated geographical area
spaced on the track, the average delay time per train is 0.5 𝑇 . per Clause 2.2 of the Standard). The relative impact of
The delay cost per train-day, xI, was based on a study by interruption cost is greater for the smaller diameter pipeline than
Lovett et al. [15], in which a model was developed to quantify for the larger diameter pipeline, and it decreases as the
delay costs as the sum of increases in operational costs, including population density increases. For LVP liquid pipelines, the
locomotive fuel costs, crew costs, emissions, as well as the costs horizontal parts of the CI/C relationship correspond to population
of delays at level road crossings. Three examples were provided density ranges in which the value of C is governed by
for unit trains, which carry a single type of goods between the environmental impact and is, therefore, independent of the
same origin and destination; manifest trains, which use railcars population density.
and locomotives on various routes based on demand; and Since each order-of-magnitude increase in the value of the
intermodal trains, which move intermodal containers between consequence indicator increases the safety class by one, CI/C is
ports and other facilities. The weighted sum of the unit costs indicative of the added conservatism applied to crossings to
associated with these categories was used as a representative mitigate the risk of traffic interruption. The results show that the
value of xI, with the weights being based on the proportion of additional conservatism is less for LVP liquid pipelines than for
train type in service, obtained from the 2019 Canadian Rail gas pipelines and that it decreases with increasing population
Trends report [16]. The results are shown in Table 4. density.

5 © 2022 by ASME
TABLE 5: INTERRUPTION COSTS FOR ROADS AND RAILWAYS WITH INFLATION AND CURRENCY CONVERSION
Increase in C due
Type Interruption Cost (USD) Unit Year Inflation Multiplier USD-CAD Conversion Interruption Cost (CAD) to Interruption
Cost
Major Road 848,000 $/lane 2008 1.24 1.27 1,335,000 65/lane
Minor Road 353,000 $/lane 2008 1.24 1.27 556,000 27/lane
Railway 393,000 $ 2015 1.12 1.27 559,000 27

100000 To conservatively capture all activities within 7 m on either side


NPS 4, at 6 MPa - 4 lane Major Road
NPS 4, at 6 MPa - Railway
NPS 4, at 6 MPa - 2 lane Minor Road
NPS 24, at 16 MPa - 4 lane Major Road
of the road, which is the pipe length required to comply with the
NPS 24, at 16 MPa - 2 lane Minor Road NPS 24, at 16 MPa - Railway design requirements for crossings, all activities within 10 m of
10000
the road or railway were included. The total length of pipe
Consequence Indicator Adjustment Factor

considered included the length under the road or railway plus


1000 10 m on each side. The hit rates were calculated from the activity
rates using a fault tree model that represents the various
100
protective and preventive measures typically implemented for
pipe under roads and railways [2], and a cover depth of 1.2 m for
roads and 2.0 m for railways. The results are summarized in
10
Table 6. The baseline hit rate was used in developing the hoop
stress factors for general buried pipe [2]. Columns highlighted in
1 grey are based on a single data point, and could not be used by
0.01 0.1 1
Surrounding Population Density (People/Ha)
10 100
themselves in quantifying the hit rates.
Based on the above information, one activity rate was used
a) Natural Gas for class locations 1 and 2, and another for class locations 3 and
100
NPS 4, at 6 MPa - 4 lane Major Road NPS 4, at 6 MPa - 2 lane Minor Road
4. Since the class location is not used in the risk-based approach,
NPS 4, at 6 MPa - Railway
NPS 24, at 16 MPa - 2 lane Minor Road
NPS 24, at 16 MPa - 4 lane Major Road
NPS 24, at 16 MPa - Railway
selection of an appropriate hit rate is based on the population
density corresponding to the boundary between class locations 2
Conseqeunce Indicator Adjustment Factor

and 3 (1.8 people/ha based on 45 dwelling units in a 400 by


1,600 m assessment area and an average of 2.5 people per
dwelling). Table 7 shows the resulting hit rates as absolute values
10 and as ratios, rh, to the default hit rate of 0.004 per km-year. All
rh values were rounded up to reasonable whole numbers.

5.2 Application
The risk of failure due to equipment impact under a road or
railway can be calculated from the hit rate as follows:
1
0.01 0.1 1
Surrounding Population Density (People/Ha)
10 100
𝑅 = 𝑟 ×𝑝 × 𝐶 = 𝑝 × (𝑟 × 𝐶 ) (8)
b) Low-flammability LVP Liquid where RR is the risk of failure for road or railway, rh is the
hit rate adjustment factor (see Table 7), pf is the probability of
FIGURE 3: IMPACT OF TRAFFIC INTERRUPTION failure due to equipment impact for ‘general’ pipe, and CI is the
ADJUSTMENT ON CI/C FOR ROADS AND RAILWAYS consequence indicator for the road or railway (Equation (5)).
Equation (8) indicates that the hit rate adjustment factor can be
5. EQUIPMENT IMPACT HIT RATES accounted for by increasing the consequence indicator by a
5.1 Approach factor of rh. The increase in the value of C will result in a
Construction activity rates were quantified using data on the potential increase in the safety class and a commensurate
number of one-call requests related to construction activities reduction in the hoop stress factor.
near roads and railways. Locate calls within the pipeline right-
of-way, which has an average width of 20 m [17], were included.

6 © 2022 by ASME
TABLE 6 ACTIVITY AND HIT RATES FOR PIPE UNDER ROADS AND RAILWAYS
Type Road Railway
Class location 1 2 3 4 1 2 3
Number of segments/crossings 17,361 727 733 8 899 52 20
Length of segments/crossings (m) 637,621 25,832 25,847 234 39,556 2,288 880
Number of one-calls in 5 years 859 49 184 1 17 1 27
Activity rate (/km-year) 0.27 0.38 1.42 0.85 0.09 0.09 6.14
Hit rate (hits/km-year) 0.010 0.014 0.057 0.034 0.001 0.001 0.088
Baseline hit rate (/km-year) 0.004 0.004 0.004 0.004 0.004 0.004 0.004
Ratio to baseline rate 2.5 3.5 14.2 8.5 0.3 0.3 22.1

TABLE 7: HIT RATES AND VALUE OF rH USED IN THE because CSA Z662 includes class-location-based reductions in
RISK-BASED DESIGN the hoop stress factors for gas pipeline crossings but not for LVP
Population Density (people/ha) pipeline crossings.
Category
<1.8 ≥1.8
Road 0.010 (3) 0.057 (15)
Railway 0.001 (1) 0.088 (20)

6. COMPARISON WITH CURRENT CROSSING


DESIGNS
Figures 4 through 7 show a comparison of the equivalent
hoop stress factors (as calculated from the final design wall
thickness) for uncased crossings designed using the risk-based
approach (coloured bars) and the current CSA Z662 approach
(non-coloured bars). The minimum wall thickness requirements
in Table 4.5 and the maximum D/t for railways in Tables 4.10
and 4.11 of the Standard were included for both approaches, so
that the only difference between the two designs is the
replacement of the pressure design approach in Clause 4.3.5 of
the Standard with the risk-based approach. The comparisons are
provided for various diameter-pressure combinations for natural
gas and LVP liquid pipelines not affecting a DGA in class
locations 1 and 3.
The figures show that the risk-based approach results in
lower hoop stress factors (i.e. thicker walls) for small diameter,
low pressure pipelines, and higher hoop stress factors (i.e.
thinner walls) for large diameter, high pressure pipelines. These
differences result from considering the equipment impact as a
design condition and are explained by the resistance to
equipment impact being more sensitive to the wall thickness than FIGURE 4: HOOP STRESS FACTOR FOR
the hoop stress. Since wall thickness in the current Standard is NATURAL GAS IN CLASS LOCATION 1
based entirely on the hoop stress, the risk-based approach
requires wall thickness increases for smaller diameter, low
pressure pipelines to attain adequate resistance to equipment
impact. The risk-based and CSA designs are similar for railways
because most railway designs are governed by the surface load
check (Table 4.11 in the Standard) in both approaches. The
Standard does not include a surface load check for roads. The
relative wall thickness increases over CSA Z662 designs for road
crossings are smaller for gas pipeline than for LVP pipelines

7 © 2022 by ASME
FIGURE 5: HOOP STRESS FACTOR FOR FIGURE 7: HOOP STRESS FACTOR FOR LVP
NATURAL GAS IN CLASS LOCATION 3 LIQUID IN CLASS LOCATION 3

7. SURFACE LOAD CHECK


The Canadian Energy Pipeline Association (CEPA) surface
load calculator [18] was used to estimate the pipe combined
stresses due to traffic loading from vehicle loads, internal
pressure, pipe properties, and soil properties. The CEPA
calculator determines the maximum stress directly under the
loaded area, which is the location of maximum stress for a single
wheel. To account for multiple wheels, the global bending
moment calculation was updated based on the approach
described in Zhang et al. [19], in which the maximum stress is
defined as a maximum of the stresses calculated at different
locations inside and outside the footprint of the loaded area.
The Von Mises criterion was used to calculate the combined
stresses on the inner and outer diameter at the top and bottom of
the pipe, and the maximum of the four combined stress values
was checked against a stress limit of 90% SMYS as specified in
Clause 4 of the Standard.
The input parameters used in the model are as follows:
• Design Parameters. Conservative values were used,
including SMYS, the wall thickness with a 95% probability
of being exceeded (based on a Normal distribution with a
mean of 1.01, the nominal wall thickness, and a coefficient
of variation of 1%), and the worst of two cases representing
unpressurized pipe and maximum operating pressure. The
nominal pipe diameter was used because diameter has
FIGURE 6: HOOP STRESS FACTOR FOR LVP negligible variability.
LIQUID IN CLASS LOCATION 1

8 © 2022 by ASME
• Vehicle Loading. The AASHTO HS20 truck loading (three 8. CONCLUSION
axles and a total load of 225 kN) was used [20,21]. The process developed to calculate the safety class for pipe
• Soil Parameters. The soil parameters used in the model are under roads or railways involves adjusting the population density
summarized in Table 8. for general pipe to account for road users (not applicable to
railways). The resulting population density is then used to
The check was performed for perpendicular two-lane minor calculate a consequence indicator (C), which is further adjusted
and four-lane major road crossings for 12,560 design cases to account for traffic interruption costs and construction activity
representing a wide range of pressure, diameter, SMYS, and rates. The final value of C is used to calculate the safety class
class location combinations [2]. Each case was analyzed for a and design the pipe as described in Nessim et al. [1,2].
natural gas pipeline and a low-flammability LVP liquid pipeline To ensure that the new approach is simple and based on
in a non-DGA. Of the 12,560 cases, the surface load check information that is readily available to designers, the various
resulted in maximum stresses exceeding 90% SMYS in 191 adjustment factors are based on generic values representing
cases (1.5%). All 191 cases were grade 241 for NPS 16 and average conditions. In implementing the approach in the
greater or grade 290 for NPS 36 and greater, which are Standard, allowance is made for designers to refine their designs
combinations that are highly unlikely to be used. It was further based on more site-specific information. Recognizing that the
found that imposing an upper bound of 0.75 on the hoop stress location factors in the Standard are currently relied upon by
factor results in all design cases meeting the surface load check. designers to provide resistance against normal traffic loads for
Considering the conservative nature of the model and all roads, the designs resulting from the risk-based approach were
input parameters used in the check, these results suggest that the checked to ensure adequacy under traffic loads.
risk-based approach produces wall thicknesses that are adequate Comparative studies show that the risk-based approach is
to resist normal surface loads for roads. However, any concern generally more conservative than the current version of the
regarding the few unlikely diameter/grade combinations that do Standard for most road crossings, except those involving large-
not meet the surface load check can be eliminated by imposing diameter and high-pressure pipelines. The added conservatism is
an upper bound of 0.75 on the hoop stress factor. smaller for natural gas pipelines than LVP pipelines because the
Standard does not currently require reduced hoop stress factors
TABLE 8: SURFACE LOAD MODEL SOIL PARAMETER for LVP pipeline crossings. Similar trends also apply for railway
INPUTS crossings, except that the difference between the two approaches
Input is smaller than for roads because many railway designs are
Value Source Comments governed by explicit surface load-based requirements in the
Variable
Standard that are unique to railway crossings.
Temperature CSA Z662 As specified in CSA Z662 for
55 °C
Differential Table 4.11 railways. ACKNOWLEDGEMENTS
This work is the result of a joint industry project sponsored
Representative value based by ATCO Gas and Pipelines Ltd., the Canada Energy Regulator,
2,000
Soil Density CEPA [22] on CEPA report and C-FER
kg/m3
experience.
Enbridge Employee Services Canada Inc., TC Energy, and Trans
Mountain Pipeline LP. The valuable guidance provided by the
Accounts for bedding
material support to the pipe.
project Steering Committee, consisting of Scott Ironside (Chair),
Bedding Warman et
Angle
30°
al. [23] Curtis Munier, Dongliang Lu, Joe Paviglianiti, and Lorena Cala-
Conservative value for open
trench installation. Phillips, is gratefully acknowledged

CEPA user Accounts for dynamic effects REFERENCES


Impact Factor 1.5
guide [22] of traffic loading.
[1] Nessim, Maher, Stephens, Mark, and Yue, Howard. “A Risk-
Accounts for soil resistance based Safety Class System for Onshore Pipelines.” Proceedings
CEPA to pipe ovalization. of the International Pipeline Conference. IPC2022-87099.
Modulus of 4,826.5
calculator
Soil Reaction kPa
[18] Conservative for fine-grained Calgary, AB, September 26-30, 2022.
soil at 90% compaction. [2] Adianto, Riski, Nessim, Maher. “Risk-based Hoop Stress
Used to determine the Factors for Pressure Design.” Proceedings of the International
Modulus of
0.03 Zhang et al. stiffness of soil springs. Pipeline Conference. IPC2022-86815. Calgary, AB, September
the
N/mm3 [19] For backfill consisting of 26-20, 2022.
Foundation
sand/gravel mixture. [3] Nessim, Maher, Stephens, Mark and Yue, Howard. Risk-
1.2 m based Safety Class System for CSA standard Z662. Technical
Depth of (road) Minimum requirements per Report No. M256. C-FER Technologies, Edmonton, AB. 2021.
CSA Z662
Cover 2.0 m Clause 4.11 of CSA Z662. [4] Stantec Consulting Ltd. Provincial Highway Service
(rail) Classification. Stantec Consulting, Edmonton, AB (2007).
[5] City of Toronto. “Attachment 1: Road classification criteria
(adopted by City Council on February 29, March 1 & 2, 2000).”

9 © 2022 by ASME
Road Classification Amendments. City of Toronto, Toronto, ON [21]American Water Works Association. Steel Pipe – A Guide for
(2018). p.4. Design and Installation – Manual of Water Supply Practices. 5th
[6] Gill, Ted. “Development of Policy Papers for Phase Two of ed. American Water Works Association, Denver, CO (2017).
the Transportation Master Plan for the City of Hamilton: Road [22] Van Auker, Mark and Francini, Bob. “Canadian Energy
Classification Policy Paper.” City of Hamilton, Hamilton, ON. Pipeline Association (CEPA) Surface Loading Calculator User
2005. Manual.” Technical Report No.: 14-017. Kiefner and Associates,
[7] Alberta Transportation. “Chapter A: Basic Design Columbus, OH. 2014.
Principles.” Highway Geometric Design Guide. Alberta [23] Warman, David J., Hart, James D. and Francini, Robert B.
Transportation, Edmonton, AB (2018). Development of a Pipeline Surface Loading Screening Process
[8] “Average Vehicle Occupancy Factors for Computing Travel and Assessment of Surface Load Dispersing Methods. Technical
Time Reliability Measures and Total Peak Hour Excessive Delay Report No.: 05-44R1. Kiefner and Associations, Worthington,
Metrics.” Federal Highway Administration Report. Vol. 18 OH. 2009.
No.112.
[9] Treasury Board of Canada Secretariat. Canadian Cost-
Benefit Analysis Guide: Regulatory Proposals. Her Majesty the
Queen in Right of Canada, Ottawa, ON (2007).
[10] Gillett, Jessica C. “Monetizing Truck Freight and the Cost
of Delay for Major Truck Routes in Georgia.” Thesis. Georgia
Institute of Technology, Atlanta, GA. 2011.
[11] Miao, Qing, Wange, Bruce X. and Adams, Teresa M.
“Assessing the Value of Delay to Truckers and Carriers.”
Technical Report No. 09-00-45. University Transportation
Center for Mobility, College Station, TX. 2011.
[12] Buyck, Sofie C.P. “Economic Loss Estimation Along
Transportation Corridors: A Method to Estimate the Economic
Loss of a Disruption of a Lifeline as a Result of an Unscheduled
Event.” University of Twente, Enshede, The Netherlands. 2008.
[13] National Cooperative Highway Research Program.
“Methodologies to Estimate the Economic Impacts of
Disruptions to the Goods Movement System.” Technical Report
No. 732. Transportation Research Board, Washington, DC. 2012.
[14] Nessim, Maher, Adianto, Riski, Nathoo, Hafeez and
Abdulhameed, Diana. “Risk-based Pressure Design Approach
for CSA Standard Z662.” Technical Report No. M256. C-FER
Technologies, Edmonton, AB. 2021.
[15] Lovett, Alexander H., Dick, C. Tyler and Barkan,
Christopher P. L. “Determining Freight Train Delay Costs on
Railroad Lines in North America.” Proceedings of the
International Association of Railway Operations Research 6th
International Conference on Railway Operations Modelling and
Analysis. Tokyo, Japan, March 2015.
[16] Railway Association of Canada. Rail Trends 2019. 27th ed.
Railway Association of Canada, Ottawa, ON (2019).ISBN: 978-
1-927520-11-6.
[17] Farmer’s Advocate Office. “Pipelines in Alberta: What
Landowners Need to Know.” Document No: Agdex 878-4.
Alberta: Farmers’ Advocate Office of Alberta. 2018.
[18] Tools and publications. Kiefer and Associates, Columbus,
OH (2015). https://kiefner.com/news/publications/.
[19] Zhang, Fan, Branam, Nathan, Zand, Bejamin and Van
Auker, Mark. “A New Approach to Determine the Stresses in
Buried Pipes Under Surface Loading.” Proceedings of the 11th
International Pipeline Conference. IPC2016-64050. Calgary,
AB, September 26-30, 2016.
[20] American Lifelines Alliance. Guidelines for the Design of
Buried Steel Pipe. American Lifelines Alliance (2001).

10 © 2022 by ASME

You might also like