Westside Rail Tunnels |
Long highway and rail tunnels are fitted with piping systems to
convey water for fire-fighting. Fire suppression usually requires
500 to 1,000 gallons of water per minute (gpm) (32 to 63 L/s) at
valve pressures from 65 to 100 psi (448 to 690 kiloPascals). These
water supplies must be kept at 40o F (4o C)
or above to prevent freezing. In cold climates, this usually requires
insulation and heating.
In designing Portland Oregon’s Westside Light Rail Tunnel
standpipe, we utilized the primary features of the tunnel design
to eliminate the need for insulation and heating, and to reduce
the tunnel piping one size. The result was an estimated savings
of more than $1 million for 6 miles (9.6 kilometers) of pipeline.
[See also “Careful Design Results in Cost Savings on the Westside
Corridor Project” by Rick Mayes, PB Network Winter
‘93/94, p. 13-15, 41, 42.]
When using computer programs for analysis, such as we did on the
hydraulic analysis for the standpipe design, PB policy requires
that one of the following procedures be completed:
- The program must be validated as generally acceptable for the
application and the input verified manually.
- Another program must be used to verify the results independently.
Description of the System
Some of the pertinent characteristics of the Westside Light Rail
Tunnels are:
- They are two 15,400-foot-long (4694-meter-long) bored tunnels
with cross-passages
- One station is located at approximately 5,400 feet (1646 meters)
from one portal A utility shaft is located approximately 2,800
feet (853 meters) from the other portal
- They change in elevation by 427 feet (130 meters) from portal
to portal
- The station at Washington Park is the deepest in North America
at 250 feet (76.2 meters) below the surface.
Freeze protection was a major consideration because Portland has
experienced continuous days of below-freezing temperatures. We selected
a dry pipeline system that remains empty until needed, thereby eliminating
the need for insulation and heating. The disadvantage of this type
of system is that there can be a significant delay before water
is available at the hose valve. We had to determine that such a
delay would not occur because one of our major objectives was to
ensure that fire fighting would not be jeopardized.
This objective was partially accomplished by designing the system
to fill when a fire is first reported to Central Control. We had
to confirm, however, that the system could fill in less than ten
minutes, a time deemed as the minimum acceptable response by the
fire agencies. Doing so required extensive hydraulic modeling.
The hydraulic modeling provided a good example of how different
programs can be used to verify results. In this
system, two different sets of hydraulic conditions had to be analyzed—the
steady-state operational conditions (occurring when used by fire
fighters) and the transient conditions, especially the initial filling
of the system.
Operational Modeling
The operational modeling was fairly straightforward. Many software
programs are available to handle steady-state fluid flow. We used
PIPEFLO, a general-purpose hydraulic network program that uses an
iterative process of changing flows to balance pressures at nodes
in the network. PIPEFLO determines pipeline/fluid resistance by
calculating a friction factor based on fluid characteristics, piping
size and roughness. This method, the Darcy-Weisbach method, is the
general approach taken in fluid mechanics.
Because this is a fire protection system, it must comply with National
Fire Protection Association (NFPA) standards. NFPA bases its friction
loss methodology on the Hazen-Williams method, a practice developed
in civil engineering for water work. This methodology is specific
to water at 60o F (16o C), but is considered
to be sufficiently accurate for the range of temperatures found
under ambient conditions. The checker felt that the check should
be based on Hazen-Williams methodology.
We used HASS, a fire protection system program based on the methodology
preferred by NFPA. We agreed that the actual unit friction loss
under flow conditions should be calculated using NFPA methods because
NFPA defines the standard in this case. In HASS, the roughness can
be entered directly. An equivalent had to be calculated for PIPEFLO.
The analysis was being conducted in Sacramento, the checker was
in New York. Having HASS in both locations allowed us to share analysis
results and even discuss them and set up new trials over the phone.
Agreement could be reached quickly or the basis of disagreement
could be analyzed and resolved. Once the calculation basis was agreed
upon, final runs were made on PIPEFLO.
Detailed analysis of the operating characteristics showed that over
time, the pipeline hydraulics would significantly worsen due to
dry-wet corrosion. Constructing the line of stainless steel, however,
would eliminate this problem.
Fill-time Modeling
The fill-time analysis was a little more difficult. Commercial piping-analysis
software is not set up to handle unsteady flows such as the filling
of a pipe. To resolve this problem, we wrote a program in Mathcad,
a freeform math application, that determined the time necessary
to fill incremental segments of the system. These times were then
summed to determine a total. [See “Mathcad Software Review”
by Ken Harris in PB Network Winter ‘93/94, pp. 26-27.]
The checker decided to use PIPEFLO to perform the verification.
He split the system into three segments and analyzed the time it
took water to travel a given distance. The results of this analysis
were consistent with the results from Mathcad. Since the PIPEFLO
model had been verified with HASS, it could be considered an accurate
calculation basis program.
Pipe size. As part of this modeling process, different
pipe sizes were evaluated. Conventional practice for a line this
size is to use 6-inch (150 mm) piping. Our analysis showed, however,
that the time for filling 4-inch (100 mm) piping was less than 6-inch
piping because of limitations in the municipal water supply system
and the fact that the 6-inch piping contains about twice the volume
of water as the 4-inch system. The final configuration that resulted
was predominantly a 4-inch system, with the westernmost 800 feet
(24.4 meters) of the tunnel sized at 6 inches in order to maintain
the required flow and pressure conditions.
Conclusions
The overall result of our test and checking was a design that resulted
in significant capital cost savings to the client. Operational and
maintenance costs also were reduced because no heating is required
and the use of stainless steel means periodic painting will not
be required.
This effort also provided an excellent example of the value of standardizing
technical software. Not only does it allow for technical accuracy
and ease of use, but it allows users to communicate in a common
language for input, output and understanding of the results. |
