Full text of "Proceedings of the annual convention"

See other formats

HBH

A90B
6k

mSsmR

111

■

««§

ran

■Hi ■ N
ills

^^H

HSirailSlliBs

Mm*

'-'■'V*

Hi
■

HH HH
IHHmlH H

I ■ ;; Hi

^H Hi

PROCEEDINGS OF THE

AMERICAN RAILWAY ENGINEERING

ASSOCIATION

(Engineering Division, Association of American Railroads)

CONTENTS, VOLUME 64

(For detailed index, see Bulletin 579, page 739)

Bulletin 573, September-October 1962

Page

Epoxy Resins 1

Termite Control Investigation — Inspection of Specimens After 52 Months

of Exposure 19

Grading Problems Encountered During Relocation of the Santa Fe Railway's

Main Line Between Williams and Crookton, Ariz 25

Conductivity Tests of Open-Hearth-Slag Ballast 35

Rail Slippage Tests — Concrete Ties 39

Preventing Rail Failures in Track 47

Welded Railroad Bridges 57

Field Investigation of Florida East Coast Prestressed Concrete Beams 69

Field Investigation of Southern Pacific Company, Texas & Louisiana Lines,

Concrete Girder Spans 70

Train Performance Calculator 73

A Computer Simulation of Railroad CTC Operations 82

Average Tie Life — An Interpretation 95

Bulletin 574, November 1962 (Reports of Committees)

16 — Economics of Railway Location and Operation Ill

9 — Highways 131

13 — Water, Oil and Sanitation Services 139

14 — Yards and Terminals 159

20 — Contract Forms 187

25 — Waterways and Harbors 197

6 — Buildings 213

Bulletin 575, December 1962 (Reports of Committees)

8 — Masonry 223

3 — Ties and Wood Preservation 241

22 — Economics of Railway Labor 263

27 — Maintenance of Way Work Equipment 305

30 — Impact and Bridge Stresses 327

28 — Clearances 2>2>2>

Bulletin 576, January 1963 (Reports of Committees)

15 — Iron and Steel Structures 359

7— Wood Bridges and Trestles 371

11 — Engineering and Valuation Records 387

24 — Cooperative Relations with Universities 397

18— Electricity 407

Bulletin 577, February 1963 (Reports of Committees)

5— Track 419

Continuous Welded Rail 449

4— Rail 497

1 — Roadway and Ballast 543

Bulletin 579, June-July 1963

Report of the Business Meeting of the Association, March 15-16, 1963 579

Report of the Executive Secretary 706

Report of the Treasurer 725

Constitution

Tie Renewals and Cost per Mile of Maintained Track 738

Index of Proceeding, Vol. 64, 1963 739

&if

American Railway

Engineering Association— Bulletin

Vol. 63, No. 573 September-October 1962

CONTENTS

Epoxy Resins 1

Termite Control Investigation — Inspection of Specimens After

52 Months of Exposure 19

Grading Problems Encountered During Relocation of the
Santa Fe Railway's Main Line Between Williams and

Crookton, Ariz. 25

Conductivity Tests of Open-Hearth Slag Ballast 35

Rail Slippage Tests — Concrete Ties 39

Preventing Rail Failures in Main Track 47

Welded Railroad Bridges 57

Field Investigation of Florida East Coast Prestressed Con-
crete Beams 69

Field Investigation of Southern Pacific Company, Texas &

Louisiana Lines, Prestressed Concrete Girder Spans 70

Train Performance Calculator 73

A Computer Simulation of Railroad CTC Operations 82

Average Tie Life — An Interpretation 95

BOARD OF DIRECTION
1962-1963

President

C. J. Code, Assistant Chief Engineer — Staff, Pennsylvania Railroad, Philadelphia 4, Pa.

Vice Presidents
L. A. Loggins, Chief Engineer, Southern Pacific Company, Texas & Louisiana Lines,

Houston 1, Tex.
T. F. Burris, Chief Engineer System, Chesapeake & Ohio Railway, Huntington, W. Va.

Past Presidents

E. J. Brown, Chief Engineer, Burlington Lines, Chicago 6.

R. H. Beeder, Chief Engineer System, Atchison, Topeka & Santa Fe Railway, Chicago 4.

Directors

C. J. Henry, Chief Engineer, Pennsylvania Railroad, Philadelphia 4, Pa.

J. M. Trissal, Vice President and Chief Engineer, Illinois Central Railroad, Chicago 5.

W. B. Throckmorton, Chief Engineer, Chicago, Rock Island & Pacific Railroad, Chi-
cago 5.

J. A. Bunjer, Chief Engineer, Union Pacific Railroad, Omaha 2, Nebr.

J. H. Brown, Chief Engineer, St. Louis-San Francisco Railway, Springfield 2, Mo.

J. E. Eisemann, Chief Engineer, Western Lines, Atchison, Topeka & Santa Fe Rail-
way, Amarillo, Tex.

W. H. Huffman, Assistant Chief Engineer — Construction, Chicago & North Western
Railway, Chicago 6.

F. R. Smith, Chief Engineer, Union Railroad, East Pittsburgh, Pa.

W. L. Young, Chief Engineer, Norfolk & Western Railway, Roanoke 17, Va.
T. B. Hutcheson, Chief Engineer, Seaboard Air Line Railroad, Richmond 13, Va.
C. E. Defendorf, Chief Engineer, New York Central System, New York 17.
John Ayer, Jr., Vice President — Operations, Denver & Rio Grande Western Railroad,
Denver 17, Colo.

Treasurer
A. B. Hillman, Retired Chief Engineer, Belt Railway of Chicago; Chicago & Western
Indiana Railroad, Chicago 5.

Executive Secretary
Neal D. Howard, 59 East Van Buren St., Chicago 5.

Assistant Secretary
E. G. Gehrke, 59 East Van Buren St., Chicago 5.

Secretary Emeritus
Walter S. Lacher, 407 East Fuller Road, Hinsdale, 111.

Published by the American Railway Engineering Association, Monthly, January, February, March,

November and December; Bi-Monthly, June-July, and September-October, at 2211 Fordem

Avenue, Madison, Wis.; Editorial and Executive Offices,

59 Van Buren Street, Chicago 5, 111.

Second class postage paid at Madison, Wis.

Accepted for mailing at special rate of postage for in Section 1103, Act of October 3, 1917,

authorized on June 29, 1918.

Subscription $10 per annum.

Advance Report of Committee 7 — Wood Bridges and Trestles
Report on Assignment 6

Applications of Synthetic Resins and Adhesives
to Wood Bridges and Trestles

Collaborating with Committees 8 and 15

L. R. Kubacki (chairman, subcommittee), R. E. Anderson, C. E. Atwater, J. W. Brent,
T. P. Burgess, J. \V. Chambers, B. E. Daniels, R. H. Hunsinger, W. D. Keeney,
C. A. Meadows, T. K. May, D. V. Sartore, F. E. Schneider, W. D. Turner, D. L.
Walker.

This report, in four parts, is presented as information. It supersedes the report on
this assignment published in Bulletin 562, January 1961, pages 526 to 537, incl. Part 1
explains the technology of epoxy resins. Part 2 summarizes the railroad applications of
epcxy resins. Part 3 presents tentative formulations for epoxy resins. Part 4 provides
general instructions for the use of epcxy resins.

The report was prepared for the committee by F. P. Drew, research engineer struc-
tures, Association of American Railroads under the general direction of G. M. Magee,
director of engineering research, AAR.

Part 1
Technology of Epoxy Resins

1. Characteristics of Epoxy Resins

The term "epoxy" refers to a three-member ring structure containing two carbon
atoms and an oxygen atom. Materials containing an average of more than one epoxy
group per molecule are considered to be epoxy resins.

Epoxy resins are thermosetting materials which can be converted with a wide
variety of curing agents to insoluble, infusible solids. The commercial use of these
resins is based on the superiority of the epoxy resins over other thermosetting resins
in one or more of the following properties:

a. The liquid resin is convenient and easy to use in many different types of
applications.

b. Versatility in curing schedule is possible with a wide variety of curing agents.

c. The property of the cured resins may be varied within wide limits witli tin
pre per choice of curing agent and modifier.

d. No volatile materials are formed during the cure of epoxy resins.

e. Low shrinkage during cure allows for accurate reproductions.

f. The combination of hardness and toughness of cured resins is outstanding.

g. The overall strength properties such as tensile, flexure, and compression strength
are excellent.

h. Resistance to solvents and chemicals of many kinds is exceptionally k'«'od.
i. Adhesion of cured epoxy resins to most surface- i- excellent,
j. Electrical insulation properties of cured epoxy resins are very good.

Epoxy resins may be used with a wide variety of fillers to further modify and
reduce cost of epoxy resin compositions. The color of many epoxy resin systems is satis-
factory for all except the most exacting applications.

Bull. 573

2 Epoxy Resins

Several modifications of the basic epoxy resins are available. The use of these mod-
ified resins is generally directed toward a specific application, since the improvement
of one property is normally accompanied by a sacrifice in one or more of the other
properties.

Most epoxy adhesives are two-package systems which are mixed just prior to appli-
cation because of limited pot life. Depending on the curing agent and the desired curing
time, these two-package systems may be cured at room temperature or an elevated tem-
perature. Systems designed to cure at room temperature normally reach handling strength
within a few hours, but usually require several days to attain maximum strength. De-
pending on the curing agent, adhesives formulated for curing at elevated temperatures
may require a few minutes to several hours for complete cure. As with other epoxy resin
applications, the choice of resin and curing agent depends on the service requirements
and desired curing schedule.

2. Curing Agents

A curing agent, which is usually added just prior to use, is required to convert
epoxy resins to thermoset polymers. Curing agents may be poly-functional compounds
containing groups which react with the epoxide ring or they may be catalysts which
promote the self-polymerization of the resin. Some curing agents act through a com-
bination of these processes.

3. Formulation

The selection of an eopxy resin system for a particular application depends not
only on the required properties of the cured resin, but also on the desired curing sched-
ule and handling properties, such as viscosity, pot life, exotherm and toxicity. The prop-
erties of both the uncured composition and the cured product can be varied over a wide
range by the proper selection and use of resin modifiers, reactive diluents and fillers.
These materials are important components of most epoxy resin formulations and infor-
mation concerning their effect on the properties of the composition before, during and
after it is cured is necessary for the most effective use of epoxy resins.

In Part 2 of this report are listed various applications of epoxy resins on timber,
concrete and steel railroad structures. For each application a suggested formulation num-
ber is indicated. These numbers refer to Part 3 of this report wherein the composition
of each formulation is given, together with certain pertinent information such as mixing,
curing, application and physical properties.

4. Resin Modifiers

Epoxy resins are modified primarily to improve the mechanical and thermal shock-
resistant properties of the cured resin, although in some cases the main purpose of the
modifier is to reduce the cost of the resin system. In general, nonreactive plasticizers,
which are used widely with thermoplastic resins, are of little value for modifying epoxy
resins. The most satisfactoiy modifiers are materials which react either with the resin
or with the curing agent to become an integral part of the cured resin. One obvious
approach is the use of curing agents or modified resins which contribute the desired
degree of flexibility.

The modifications of epoxy resins to improve impact and thermal shock resistance
is usually accompanied by a loss of strength as well as electrical insulation properties
and chemical resistance. Since the loss of properties is most evident at elevated tem-
peratures, an epoxy resin should not be modified more than necessary to meet the
requirements for a specific application.

Epoxy Resins 3

5. Reactive Diluents

For many applications it is necessary to reduce the viscosity of epoxy resin for-
mulation and at the same time maintain as nearly as possible the original properties
of the cured resin. Since the most suitable materials for this purpose react to become
a part of a cured system, they are commonly called reactive diluents. A reactive should
contribute good viscosity reduction at low concentration, be nonreactive with the resin
under normal storage conditions, react at approximately the same rate as the resin and
have little effect on the properties of the cured resin.

The most effective reactive diluents are low-viscosity epoxy compounds. In addi-
tion to viscosity reduction they contribute better wetting and penetration and permit
a higher loading of inert filter. The degree to which the properties of the resin are
affected depends not only on the choice of diluent, but also on the amount used. The
properties most affected arc chemical and solvent resistance and strength at elevated
temperatures. Although most reactive diluents may be used in ratios of 25 or more parts
per 100 parts of resin, it is recommended that no more be used than is actually required
for a given application.

Properties such as pot life, curing schedule, and exotherm are usually altered some-
what by the presence of a diluent. Epoxy resins containing reactive diluents are available
commercially.

6. Fillers

In addition to reducing the cost of epoxy resin formulations, fillers are used to
modify properties of both the cured and uncured composition. For many applications
the success of an epoxy formulation may depend on the judicious use of fillers. Advan-
tages to be gained by the use of fillers include extended pot life, reduced exotherm,
lower shrinkage during cure, flow control, reduced coefficient of thermal expansion and
increased thermal conductivity. These effects are dependent on volume loading as well
as on the specific filler being used. With the exception of fibrous, reinforcing fillers,
properties such as tensile, flexural and compressive strength are adversely affected by
the addition of fillers.

Epoxy resins have a remarkable tolerance for fillers of almost every variety. They
should be nonreactive with both resin and curing agent and should be stable under the
conditions at which the formulation is to be cured and used. For most applications
the limiting factor on filler content is the viscosity of the uncured mixture. Reactive
diluents may be used to permit higher loadings in many applications. Handling of for-
mulations at elevated temperatures also allow a greater concentration of filler to be used.
For a given weight of filler a larger particle size gives less viscosity increase bul ilii-
advantage must be balanced by the greater settling tendency of larger parti

In general, colloidal silica fillers and fibrous fillers such as asbestos pri duce a max
imum pourable viscosity at less than 10 parts per hundred parts of resin (phr). At
the usual 200 to 325 mesh size, fillers such as silica, calcium carbonate, mica, flint,
pumice and slate ran be used at loadings of 50 to 150 phr. Atomized aluminum, alumina
and calcium silicate are used at 100 to 200 phr, and the heavier oilers such as iron and
iron oxide can be used up to .?00 phr. In some applications it is possible to use 50-mesh
sand at loadings as high as 800 phr.

The choice of filler depends largely on the application, Silica and alumina are used
extensively in electrical potting applications. Adhesives are Improved considerably with
moderate loadings of alumina and asbestos fillir. Parts which may require machining
may be loaded with powdered aluminum, calcium carbonate or caldum sili

4 Epoxy Resins

Aluminum-filled compositions are widely used in metal forming tools, and alumina,
flint and carborundum are used to improve abrasion resistance. Maximum leadings of
alumina, calcium carbonate or silica are recommended for reducing the coefficient of
thermal expansion. Aluminum powder is used to improve thermal conductivity, although
it is only marginally better than alumina in this respect. Steel wool has been suggested
for improving mechanical properties as well as thermal conductivity, but special tech-
niques are required for fillers of this type.

For best results fillers should be free of moisture and absorbed gases. Thorough
mixing can usually be accomplished with mechanical stirring at an elevated temperature,
although for some fillers grinding or milling may be necessary.

7. Handling Precautions

Prolonged or frequent skin contact of materials used in epoxy resin systems may
cause dermatitis for some individuals. The reactive diluents that are used in epoxy resins
have been found to be sources of skin irritation.

In working with epoxy resins it is important to realize that dermatitis problems
may exist even though they may not be apparent at first. Most persons can work with
these materials for some time without taking any precautions to avoid skin irritation.
A few persons under these conditions may suddenly break out with skin irritations
which will disappear when they are transferred to other types of work. Those who
return to work with epoxy resins and experience skin disorders very quickly are to be
considered sensitized and should discontinue work with the resins. Some individuals
become relatively immune to the materials and with normal contact do not experience
any difficulty.

Dermatitis can be avoided or held to a minimum by the use of proper equipment
and handling technique. The following protective measures have been effective in reduc-
ing the incidence of dermatitis:

a. Reasonable care in preventing skin contact and regular washing of hands, arms
and face with warm, soapy water.

b. Effective ventilation to prevent the accumulation of vapors of the volatile
amines and reactive diluents.

c. Storage of materials in closed containers fitted with spigots and valves.

d. Use of disposable containers for mixing the materials to keep cleaning of con-
taminated equipment to a minimum.

e. Solvents such as denatured alcohol for clean up must be used with extreme
care since contaminated solvents are a prime source of irritation.

f. Use of rubber gloves and prctective clothing is helpful but they should not be
worn after they have been contaminated.

Part 2
Applications of Epoxy Resins

The recommended formulations for the following applications are based on the
assumption that the surfaces on which they will be used are dry. For damp surface
conditions alternate formulations can be used, as indicated below:

On Dry Surfaces On Damp Surfaces

Formulation PI Formulation P2

Formulation 991-67 Formu'ation 991-76

Formulation 1101-32 Formulation 991-75

Epoxy Resins 5

While the damp-surface formulations can also be used on dry surfaces, they are
particularly designed to cure well in the presence of moisture. However, they are less
flexible and have a shorter pot life than the corresponding formulations recommended
for use on dry surfaces.

A. APPLICATIONS FOR COMMITTEE 7— WOOD BRIDGES AND TRESTLES

1. Protection of Pile Cut-Offs

Prime surfaces with formulation Pi, then apply adhesive formulation 991-67 1/16
in thick as a protective coating.

2. Hardening of Bearing Surfaces

Prime surfaces with formulation Pi, then apply adhesive formulation 991-67.
Incorporate either emery grit or graphite. The emery will produce an abrasive surface,
and the graphite will produce a hard sliding surface.

3. Repairs to Checks and Splits

Prime surfaces with formulation Pi, then, assuming that the application is for
timber impregnated with creosote, apply adhesive formulation 991-67.

4. Adhesive for Laminating Timbers

Prime surfaces with formulation Pi, then, assuming that the application is for
timber impregnated with creosote, apply adhesive formulation 991-67.

5. Adhesive for Gluing Component Parts of Trestle

Prime surfaces with formulation Pi, then, assuming that the application is for
timber impregnated with creosote, apply adhesive formulation 991-67.

6. Adhesive for Bonding Concrete Surfacing to Laminated Timber Girders

If cast-in-place concrete is to be used, coat the timber surfaces with formulation
991-61, 991-75 or P2, and cast the concrete onto the wet epoxy formulation.

If precast concrete is to be used prime both the timber and the concrete surfaces
with formulation Pi. Use formulation 991-67 to completely fill space between concrete
and timber.

7. Bonding Other Wearing Surfaces to Timber Decks

Prime surfaces with formulation Pi, then apply formulation 991-67.

8. Spraying for Fire Retardants

No suggestions.

9. Filling Rotted Cores in Piles, Caps, Etc.

The decayed wood should be removed to expose a sound substrate. Formulation
991-67 is suggested for filing the volume from which the decayed material has been
removed. If possible, first apply a prime coat of formulation Pi. Large cavities can be
partially filled with sound wood, using the adhesive for filling the balance <>t the void.

B. APPLICATION'S FOR COMMITTEE 8— MASONRY
1. Waterproofing of Concrete and Dampproofing
Backs of Abutments, Retaining Walls, Etc.

Use formulation Pi. It may be applied by brush oi spray equipment.

6 Epcxy Resins

2. Surfacing Car Washing Platforms

Use formulation 632-1255, a pigmented coating, for concrete washing platforms.
It may be applied by brush or spray equipment.

If the platform has a concrete surface that is pitted or spalled, prime the surface
with formulation Pi, then apply the epoxy floor-topping compound, formulation 1101-
32. This formula may be pigmented to any desired color, and may be applied tV to
% in thick.

If a non-skid surface is desired on the washing platforms, a brush or squeegee
coat of formulation Pi may be applied, and sand or an abrasive aggregate such as
carborundum may be strewn on top of the wet application. Following an overnight
cure at room temperature (75 F or above) the surface is ready for light traffic. If
waterproofing the surface is required, use formulation Pi.

3. Bonding New Concrete to Old Concrete

Use formulation 991-61, 991-75 or P2. Apply a brush coat onto the old concrete
and cast the fresh concrete onto the wet epoxy to effect a bond.

4. Splicing Concrete Piles

Prime surfaces with formulation Pi, then use formulation 991-67. Apply a brush
coat to both concrete surfaces prior to splicing.

5. Protection of Post-Tensioning Anchorage

If chemical and solvent resistance is of prime importance and water resistance is
not required, use Epi-Rez 285-1 primer with Epi-Rez 285-6 as an enamel finish. If
water resistance is of prime importance use Epi-Rez 285-22 primer with 632-1255 as
an enamel finish. These may be applied by brush or spray equipment.

6. Bonding for Composite Section of Concrete and Steel

Prime concrete surfaces with formulation Pi, then use formulation 991-67. Apply a
brush coat of 991-67 to both surfaces to be bonded.

7. Sealing Joints Where Masonry Stones Are Moving Under Load

Prime surfaces with formulation Pi, then apply formulation 991-67.

8. Grout for Heavy Bed Plates

Use formulation 991-67.

9. Joint Filler

Use formulation Epi-Rez 504-33.

10. Repair of Membrane Waterproofing
Use formulation Epi-Rez 242-2.

11. Fastening Rubber Pads to Concrete

Prime surfaces with formulation Pi, then apply adhesive formulation 991-67.

12. Shear Keys Between Concrete Beams

Prime surfaces with formulation Pi and then fill keyway with formulation 1101-32,
using 600 parts of sand per 100 parts of resin by weight. If a faster cure time is required
use formulation 991-75 instead of 1101-32. Use of electric heating cables or steam in
the key-way will further accelerate the cure.

Epoxy Resins 7

C. APPLICATIONS FOR COMMITTEE 15— IRON AND STEEL STRUCTURES

1. Protection Against Corrosion

If chemical and solvent resistance is of prime importance and water resistance is
not required, use Epi-Rez 285-1 primer with Epi-Rez 285-6 as an enamel finish. If water
resistance is of prime importance use Epi-Rez 285-22 primer with 632-1255 as an
enamel finish. The materials may be applied by brush or spray equipment.

These applications are two-ccat systems. If time permits only a one-coat system,
use formulation Cl. This is not considered a substitute for the two-coat system and
should be used only if time does not permit using the two-coat system.

2. Fastening Masonry Plates to Concrete

Prime concrete surfaces with formulation Pi, then apply formulation 991-67. Apply
a brush coat to both surfaces to be bonded.

3. Preventing Slippage of Bolted Joints

Use adhesive formulation 991-67. It may be applied on the contact surfaces and
on the bolt shanks.

4. Waterproofing Steel Plates in Lieu of Other Waterproofing Methods

Use Epi-Rez 285-22 and formulation 632-1255. If time permits only a one-ccat
system, use formulation Cl. This is not considered a substitute for the two-coat system
and should be used only when time does not permit using the two-coat system.

5. Tacking Welded Sub-Assemblies Prior to Welding

Use formulation 991-67. The application of heat (blow torch) to the bonded mem-
ber will speed up the rate of set of this adhesive.

6. Shipping Shop Assembled Sub-Assemblies

Use formulation 991-67. The application of heat (blow torch) to the bonded mem-
ber will speed up the rate of set of this adhesive.

7. Bonding Timber Decks and Ties to Steel Girders

Prime the timber surfaces with formulation Pi, then apply formulation 991-67 to
either the timber or the steel surfaces.

8. Bonding Elevation Blocks to Bridge Decks

Prime the timber surfaces with formulation Pi. then apply formulation 991-67 to
either the timber or the steel surfaces.

9. Sealing Hold-Down Devices and Spikes Firmly in Bridge Timber

Formulation Pi is recommended where a fluid system is needed to obtain good
penetration. If the space to be filled IS large, formulation 991-67 or 1101-32 can
be used.

10. Applying Anti-Slip Surfaces on Steel Plates,
Stairs and Other Areas for Positive Traction

Apply a squeegee coat of formulation Pi and sand, or an abrasive aggregate such
as carborundum may be strewn on top 01 the wet coat. Following an overnight cure
at room temperature (75 F or above), the surface i- ready for traffic.

Epoxy Resins

Part 3
Formulations

FORMULATION 242-2, EPOXY— TAR COATING

Formula:

Pounds

Gallons

Material

25.0

1.42

Cab-O-Sil

300.0

12.86

Silver Lake talc

433.0

44.41

Tar @ 75% solids*

361.0

38.00

Epi-Rez 242

24.0

3.31

Solvesso 1 00

1143.0

100.00

32.5

3.91

Tetraethylene penta

Catalyze:

To 100.0 lb 242-2, add 2.9 lb tetraethylene pentamine. Age 1 hr, then apply.

Constants:

Viscosity
Wt/Gal
P.V.C.

Vehicle solids
Total solids
Converter
Pot life

Resin Composition:

50% tar solids
50% epoxy solids

Thixotropic

11.43

18.4%

79.4%

85.3%

10.0% on epoxy solids

4-8 hr

* Various tars can be used; successful results were obtained with those having a melting point
of 50-60 C and a free carbon content of 10-20%.

EPI-REZ 285-1, ZINC CHROMATE— IRON OXIDE PRIMER

Formula and Manufacturing Procedure:
Pounds Gallons Material

245.0

5.70

Pure iron oxide

36.8

.79

Zinc oxide

36.8

1.26

Zinc chromate

116.0

4.80

Asbestine

116.0

5.16

Whiting

408.0

47.00

Epi-Rez 285

124.0

17.29

Xylol

136.0

18.00

Butyl cellosolve

1218.6

100.00

14.7

1.71

T.E.P.A. (tetraethyh

Reduce and Catalyze:

3:1 (by weight) + 1.2 lb T.E.P.A/100 lb 632-768

Reduction solvent for spray — 2, 1, 1 tuluol, butanol, M.E.K. (by volume)
Reduction solvent for brushing — 2, 1, 1 xylol, butyl cellosolve (by volume)
Allow to stand 1 hr before using

Epoxy Resins 9

Constants:

Viscosity

77 K.U.

Wt/Gal

12.19

Red. viscosity

spray — (after 1 hr aging) — 19-21 sec, No. 4 Ford cup

brush — (after 1 hr aging) — 24-27 sec, No. 4 Ford cup

Application

spray/brush

P.V.C.

41.4%

Total solids

65.5%

Vehicle solids

36.6%

Converter

6% T.E.P.A. (on resin solids)

Pot life

8-12 hr

EPI-REZ 285-6, GRAY ENAMEL
Formula and Manufacturing Precedure-

Materials
Rutile non-chalking titanium dioxide
Lampblack

IAF Compound X-2280 Imperial Color & Chem.
Epi-Rez 285
Xylol
Butyl cellosolve

Tetraethylene pentamine

Pounds

Gallons

100.0

2.86

3.0

.20

4.0

.18

698.0

80.00

60.5

8.41

62.7

8.35

928.2

100.00

16.8

2.04

Co.

Reduce and Catalyze:

For Spray— to 100 lb of 285-6 add 35 lb of reducing solvent and 1.8 lb of T.E.P.A.

Allow to stand 1 hr before using.
For Brush— to 100 lb of 285-6 add 20 lb of reducing solvent and 1.8 lb of T.E.P.A.

Allow to stand 1 hr before using.
Reducing Solvent Lineup — 2-1-1 parts by volume diacetone alcohol, toluol and butyl

cellosolve.

Constants:

Viscosity

75 K.U.

Wt/gal

9.28

Red. viscosity

spray (1 hr aging) 19-22 sec. No. 4 Ford cup

brush (1 hr aging) 54-57 sec. No. 4 Ford cup

Application

brush or spray

P.V.C.

6.8%

Vehicle solids

51.6%

Total solids

56.7%

Catalyst

4% on resin solids

Pot life

8-12 hr

EPI-REZ 285-22, RUST INHIBITIVE PRIMER
Formula and Manufacturing Procedure:

Pounds

Gallons

Part

Material

600.0

17.50

M-50 lead silica chromatc

4.0

.50

Soya lecithin

261.0

30.00

Epi-Rez 285

71.7

10.00

Xylol

112.3

15.00

Butyl cellosolve

15.8

2.00

Diacetone alcohol

1064.8

75.00

National Lead Co

1(1

Epoxy Resins

Viscosity— 72 K.U.
Wt/gal— 14.18 lb

Part B
171.5 22.00

23.7 3.00

195.2

25 mi

Versamid 401
Diacetone alcohol

Viscosity — 70 K.U.
Wt/gal— 7.82

.General Mills

Composite Blend

1064.8 75.00 Part A

195.2 25.00 Part B

1260.0

100.00

Constants:

Viscosity

70 K.U.

Wt/gal

12.60 lb

P.V.C.

39.4%

Vehicle solids

39.5%

Total solids

68.5%

Application

brush after aging

blend approximately 1 hr

Pot life

8-24 hr

FORMULATION 504-33— JOINT FILLER

Resin Portion:

Part A

Same Formula

Based on
100 Parts by

Pounds

Gallons

Material

Weight of Resin

56.8

6.14

Epi-Rez 504

100

8.5

0.57

Bentone 34

29.5

1.34

Surfex MM

5.2

0.31

Asbestos 7RF6

100.0

8.36

Converter: Part B

73.8

9.36

Epi-Cure 854

130

11.3

0.75

Bentone 34

45.4

2.06

Surfex MM

8.5

0.50

Asbestos 7RF6

139.0

12.67

Constants:

Viscosity

Paste

Wt/gal

11.36 lb

Pot life

8 hr

for 1 gal. at 77 F

Mixing:

Part A — Combine Epi-Rez 504, Bentone 34, Surfex MM and Asbestos 7RF6. Blend
thoroughly. Pass resin composition through a three-roll paint mill. (Color is light gray).

Part B — Combine Epi-Cure 854, Bentone 34, Surfex MM and Asbestos 7RF6. Blend
thoroughly. Pass converter composition through a three-roll paint mill. (Color is buff).

When ready to use combine Part A and Part B and mix thoroughly until a uniform
color is obtained

Epoxy Resins

Application:

Caulking compound may be applied by putty knife, spatula, or caulking gun.

Curing Schedule:

Gel time, 16 hr at 74 F for thin sections. At 74 F the compound continues to cure
lor several days. At temperatures below 70 F this composition will cure only very slowly.

Physical Data:

This material develops a hardness of Shore A 70 in five days and will not harden
appreciably on further aging. The water resistance and caustic resistance is quite good,
although immersion in strong solvents and acids is not recommended.

Material Sources:

Epi-Rez 504
Bentone 34
Surfex MM
Asbestos 7RF6
Epi-Cure 854

(Jones-Dabney Company)
(National Lead Company)
(Diamond Alkali Company)
(Johns-Manville Asbestos Fibre Division)
(Jones-Dabney Company)

FORMULATION 632-1255
Description and Use:

Epi-Rez 285 — Versamid 401 gray enamal

Formula and Manufacturing Procedure:

Pounds

Gallons

(Part A)

Material

100.0

2.86

Rutile non-chalking TiOa

3.0

.20

Lamp black

4.0

.18

IAFX-2280

350.0

40.14

Epi-Rez 285

98.0

13.00

Butyl cellosolve

81.0

10.29

Diacetone alcohol
Viscosity— 61 KU,

636.0

66.67

wt/gal-

(Part B)

233.0

29.87

Versamid 401**

27.0

3.46

Diacetone alcohol

260.00

33.33

Composite Blend***
636.0 66.67

260.0 33.33

896.0

100.00

Viscosity— 89 KU, wt/gal— 7.80

Part A
Part B

Resin Composition

60 — epoxy solids
40 — Versamid solids

*' If Versamid 100 is preferred, then use 60% by weight (if the Versamid
'• Allow mixture to age 1 hr before using.

Constants:

Viscosity 72 KU

Wt/cal 8.96

P.V.C. 7.5%

Vehicle solids 44.4%

Total solids 51.0%

Pot life 24-^8 hr

Epoxy Resins

FORMULATION 991-

Resin Portion

Same Formula Based on

Pounds

45.35

54.42

0.23

Gallons
4.67
2.46
0.01

Material
Epi-Rez 509
Surfex MM
Lamp black S-100

100 Parts by Weight of Resin
100
120

0.5

100.00

7.14

Converter :

22.67

4.53

36.28

2.14
0.56
1.64

Thiokol LP-3
DMP-30
Surfex MM

50
10
80

Constants:

Wt/gal
Viscositv
Pot life'

14.24 lb, including converter

brushing paste

45 min-1 lb batch, at 72 F

Mixing:

Blend Epi-Rez 509, Surfex MM, and lamp black. Stir thoroughly. Pass combina-
tion over a three-roll mill two passes for good dispersion. Blend Thickol LP-3, DMP-
30, and Surfex MM, and stir thoroughly. Pass combination over a three-roll mill two
passes for a good dispersion. When ready to use combine the resin mixture with con-
verter mixture and blend thoroughly.

Application:

Apply this adhesive to the old concrete surface evenly with a stiff brush to a film
thickness of at least 15 mils. Apply immediately fresh concrete on top of the adhesive.
No more should be mixed at one time than can be used within the limits of the pot life
of the adhesive.

Suggested Uses:

For bonding old to new concrete, under water; for bonding old to new concrete
in air; for sealing cracks on horizontal concrete surfaces.

Cure:

Time required for curing under water is approximately 14 days. Cure time at room
temperature is over night.

Material Sources:

Epi-Rez 509
Surfex MM
Lamp black S-
DMP-30
Thiokol LP-3

(Jones-Dabney Company)
(Diamond Alkali Company)
100 (Monsanto Chemical Company)

(Rohm and Haas Company)
(Thiokol Chemical Company)

FORMULATION 991-67, GENERAL PURPOSE ADHESIVE

Resin Portion:

Pounds Gallons

Material

Same Formula Based on
100 Parts by Weight of Resin

64.51

25.81

9.68

6.65
1.21
0.31

Epi-Rez 509
Asbestos 7-TF-l
Alumina T-60

100
40
15

100.00

8.17

Epoxy Resins 13

Converter:

32.26 4.08 Epi-Cure 855 50

6.45 0.37 Asbestos 7-TF-l 10

6.45 0.21 Alumina T-60 10

Constants:

Wt/gal 11.31 lb, including converter portion

Pot life 2]/2 hr at 77 F 1-lb batch

Viscosity Brookfield, No. 6 spindle, at 77 F

RPM Cps

5 56,000

10 34,000

20 22,500

Mixing:

Blend the Epi-Rez 509, 7-TF-l Asbestos, and T-60 Alumina to a smooth texture.
Blend the Epi-Cure 855, 7-TF-l Asbestos, and T-60 Alumina to a smooth texture.
When ready to use combine the resin and converter compositions and blend
thoroughly.

Curing Schedule:

This adhesive will reach its maximum strength at 77 F after 2 days. At 250 F
the maximum strength is obtained in 10 min.

Physical Properties:

The following shear strengths were obtained after five days at room temperature:

1. Creosoted wood to creosoted wood 1,610 psi in shear

Wood failure on all tests.

2. Creosoted wood to steel 1,744 psi in shear

Wood failure on all tests.

3. Steel to steel 2,793 psi in shear

4. Aluminum to aluminum 2,130 psi in shear

5. Rubber to concrete 1 20 psi

The elongation of the rubber was 285%, at ultimate shear.

The adhesive was applied approximately 5 mils thick on each surface to be bonded.
Only sufficient pressure to achieve contact was applied. All samples were allowed to cure
at room temperature for a ptriod of five days prior to testing.

Material Sources:

Epi-Rez 509 ( Jones-Dabney Company)

Asbestos 7-TF-l (Johns-Manville Asbestos Fibre Division)

Alumina T-60 (Aluminum Company of America)

Epi-Cure 855 (Jones-Dabney Company)

FORMULATION 991-75

Resin Portion: Same Formula Based on

Pounds Gallons Material 100 Paris by Weight of Resin

100.00 10.31 Epi-Rez 509 100

Converter:

35.00 4.27 Epi-Cure 872 35

Filler:

(Variable) (Variable) Clean, dry, well graded masonry sand

The amount of sand to use will depend on the gradation of the taod and the
degree of workability desired. Mix trial batches to determine best proportions.

Epoxy Resins

Constants:

Unfilled Filled

Pot life at 77 F in 1-lb batch y2 hr Ya, hr

Pot life in thin films at 77 F 2 hr 2 hr

FORMULATION 991-76

Resin Portion:

Pounds

Gallons

Material

Part

71.42

7.36

Epi-Rez 509

100

14.29

0.47

Alumina T-60

14.29

0.67

Asbestos 7-TF-l

100.00 8.50 140

Converter :

25.00 3.05 Epi-Cure 872 35

15.00 0.50 Alumina T60 21

15.00 0.70 Asbestos 7-TF-l 21

55.00 4.25 77

Constants:

Viscosity at 1 rpm, 77 F (cps) 84,000

Viscosity at 5 rpm, 77 F (cps) 35,200

Pot life at 77 F for a 1-pint batch (min) 35

Material Sources:

Epi-Rez 509 (Jones-Dabney Company)

Alumina T-60 (Aluminum Company of America)

Asbestos 7-TF-l ( Johns-Manville Asbestos Fibre Division)

Epi-Cure 872 (Jones-Dabney Company)

FORMULATION 1101-32

Same Formula Based on
Pounds Gallons Material 100 Parts by Weight of Resin

100 10.31 Epi-Rez 509 100

Converter :

50 6.33 Epi-Cure 855 50

Filler:

(Variable) (Variable) Clean, dry, well graded masonry sand

The amount of sand to use will depend on the gradation of the sand and the degree
of workability desired. Mix trial batches to determine best proportions.

Constants:

Pot life (filled) 3-4 hr

Pot life (unfilled) \y2-2 hr

mixing 1,900 cps at 77 F on Brookfield Spindle

No. 3 at 20 rpm 33% of liquid

Converter

Wt/gal (unfilled) 9.0 lb

Mixing:

Blend resin, converter, and sand, in that order.

Epoxy Resins

Curing Schedule:

Light traffic — overnight at 77 F
Heavy traffic — 3 days at 77 F

Material Sources:

Epi-Rez 509 (Jones-Dabney Company)

Epi-Cure 855 (Jones-Dabney Company)

Crystal silica sand (Ottawa Silica Company)

FORMULATION Pi

\<\ >i.\ Portion:

Pounds Gallons
100 10.31

Converter:

6.33

Material
Epi-Rez 509

Epi-Cure 855

Same formula Based on

100 Parts by Weight of Resin

100

Pot life: 100-g batch at 77 F = 4J^ hr.

FORMULATION P2

Resin Portion:

Pounds Gallon*
100 10.31

Material
Epi-Rez 509

Epi-Cure 872
Pot life: 100-g batch at 77 F = 38 min

Convert kr :
35

4.27

Same Formula Based on

100 Parts by Weight of Resin

100

Resin Portion: Part .1
Pounds Gallons

100

10.31

0.45

FORMULATION Cl

Material

Epi-Rez 509
Cabosil M5

Same Formula Based on
100 Parts by Weight of Resin

100

Converter Portion: Part B

35 4.27 Epi-Cure --7-

Mixing:

1. Using separate containers, measure out 108 parts by weight of the resin portion
( white paste marked "A") and 35 parts of the converter portion (brown liquid marked
"B").

2. Put the 108 part- ol "A" and 35 parts of "BM together in a third container.

3. Blend "A" and "B" together until the mix has a uniform color with oo streaks
Be sure that no unmixed resin adheres to the side of the container. It should take from
3 to 5 min to obtain thorough blending.

4. The epoxy mixture should now be applied without delay to the surfaces to be
coated. An inexpensive 4-in brush i- recommended. Brush it on in a thick even coal
The coverage should be approximately 100 sq ft per gal on a reasonably smooth sur
face, less coverage will be realized over areas containing rivel head- or deep pits. It the
rate of application i- dower than anticipated, the hardening time for the batch (pot
life) can be extended by emptying tht containei ol epoxj along the surface to be

and then brushing it out for the required covei

16 Epoxy Resins

Curing Time:

When the epoxy is spread out in a thin film, the internal heat of the reaction is
rapidly dissipated into the air. Because of this, the curing time of the epoxy ccating
is dependent on the surrounding temperature; below 60 F it will not harden, between
60 and 70 F it will harden in a few hours and obtain maximum strength in three to
four days. If a second or touch up coat is necessary, it can be applied as soon as the
first coat is hard enough so that the second application will not disturb it.

Part 4
General Instructions for Use of Epoxy Resins

1. Surface Preparation

For Application to Timber

Untreated timber will normally not require any special surface treatment. Surface
brushing or blowing with high-pressure air will remove dirt, dust and loose particles.
Make certain the air compressor is equipped with an oil-water trap. Oil and grease
should be removed with a solvent. The surface should preferably be dry.

Timber treated with a preservative such as creosote, coal-tar or petroleum oil must
have the preservative removed from the surface before applying the epoxy resin. Loose
oil must be scraped off and the surface washed with a solvent. A final rubbing with
clean cloths should then produce a suitable surface. Treated timber that has weathered
a number of years and has no surface oil present can be cleaned as indicated for
untreated timber.

For Application to Concrete

Concrete has a natural roughness that is favorable to the acceptance of epoxy resin.
New concrete and old concrete that is still sound should be well brushed and air blown
with high presssure air. Any oil or grease on the surface should be removed with a
solvent and the surface wiped with clean cloths. If the surface has been treated with a
curing membrane, oil, silicone, etc., such materials must be removed, sand blasting may
be necessary. Laittance must also be removed. Loose material on old concrete that has
surface deterioration must be removed to sound concrete. All pulverized and powdered
concrete must then be removed from cracks and cavities. A high-pressure water jet can
be used for this, followed by blowing with high-pressure air supplied by a compressor
with an oil-water trap. Allow to dry as necessary for the type of epoxy resin to be used.

For Application to Steel

Until various surface treatments have been evaluated it is recommended that steel
surfaces be sand blasted to receive the epoxy resin. If sand blasting is not practical, the
surfaces should be thoroughly scraped, wire brushed to remove loose mill scale and
cleaned with a solvent to remove all dirt, oil, grease and other materials that would
prevent the epoxy resin from bonding to the base metal. Wipe with clean cloths. This
is not a substitute for sand blasting, but may in some instances produce satisfac-
tory bond.

For Application to Rubber

Wash surfaces thoroughly with methyl ethyl ketone to remove any mold-release
agent.

E poxy Resins 17

2. Mixing

The manufacture's instructions should be read and followed. Some mixtures require
a specific time for aging before application, others do not. Note also the specified pot
life, as this can be a guide as to how fast the mixture will set up.

Mix small quantities at a time. Usually 1 to 3 gal can be most conveniently handled.
Use separate containers and measure parts, by volume or weight, of resin, converter
and fillers as specified in Part 3, Formulations.

Thorough blending of all ingredients is essential. Blend the resin and converter
before adding the filler. The mixing container should be of larger capacity than the
volume to be mixed. For example, for a 1-gal batch, use a 3-gal container. Mix thor-
oughly so that all components are blended uniformly from top to bottom as well as to
the sides of the container.

If the resin portion and the converter portion are furnished in contrasting colors,
the time of mixing can be determined by the time required to attain a uniform color.
If color contrast is not apparent, a 1-gal mix should be hand mixed with a spatula for
5 min or mixed with an electric paint-type mixer for 2 min.

Care should be taken to not contaminate the resin and converter by reusing mixing
tools covered with other ingredients.

The chemical reaction between resin and converter generates heat. When large quan-
ties are mixed, this heat cannot be dissipated through the sides of the container or into
the air and will accelerate the setting of the mixture. Pot life is thus influenced by the
volume of the mixture and the degree with which its heat of reaction can be dissipated.
Spread out in a thin film, it will set up slower than if in a deep container.

Advance Report of Committee 3 — Ties and Wood Preservation*
R. B. Radkey, Chairman

Termite Control Investigation — Inspection of Specimens
After 52 Months of Exposure

Introduction

This report summarizes the results of an inspection of treated specimens and un-
treated control specimens after about 52 months of exposure in a test plot established
in the Austin Cary Memorial Forest of the University of Florida near Gainesville, Fla.,
to determine the most effect preservatives and preservative retentions to reduce the rate
of decay and termite attack in oak, pine and fir wood.

The initial installation, consisting of 1296 treated specimens and 30 untreated con-
trol specimens, was made in November 1957. A total of 9 different preservatives was
used, with 3 different retentions for each preservative. A report describing the initial
installation is published in the 1959 Proceedings, Vol. 60, page 131. It covers in detail
the treatment of the specimens, chemical analyses of the preservatives, and method of
installation. A supplemental installation of 576 specimens treated with 4 additional pre-
servatives, and 15 untreated control specimens, was made in November 1959. A detailed
report describing this installation is published in the 1961 Proceedings, Vol. 62, page 95.

The results of an inspection of the specimens after 40 months of exposure are set
forth in AAR Research Department report ER-16 and in the 1962 Proceedings, Vol. 63,
page 53. The general conclusion was that only those specimens treated with coal tar
creosote completely resisted both decay and termite attack during that interval.

The investigation is being conducted under the general direction of G. M. Magee,
director of engineering research, Research Department, AAR. The conduct of the inves-
tigation and preparation of the report were in charge of E. J. Ruble, executive research
engineer, research staff, AAR. The inspection of the specimens was conducted by Dr.
J. B. Huffman of the University of Florida, R. B. Radkey, chairman of Committee 3 —
Ties and Wood Preservation, F. J. Fudge and W. L. Kahler, members of Committee 3,
and I. A. Eaton, research staff, AAR. Funds for the investigation are being provided
by the AAR.

Grading

The system of specimen grading for both decay and termite attack used during the
inspection corresponds to that recommended by the American Wood Preservers Associa-
tion and is fully described in the previously mentioned reports covering the inspection
after 40 months of exposure.

Field Inspections

The inspection of the specimens was conducted in the same manner as described in
the above mentioned reports. It involved the removal of each 2- by 4- by 18-in speci-
men, which is buried in the ground for half its length, cleaning off the dirt and sand
with a spatula and then determining its grade for both decay and termite attack. The
specimen was then replaced in its original position. The decay and termite grades are
recorded on "Field Inspection Data Sheets" an example of which is shown in the above
mentioned reports.

• An abstract of Report ER-23 issued by the Research Department. AAR. Copies of the full
report can be obtained from the director of engineering revarch, AAR, 3140 S. Federal St., Chicago 16

20 Termite Control Investigation

Analysis of Field Inspection Data

The data on the decay and termite attack grades, as recorded on the ''Field Inspec-
tion Data Sheets" were summarized and recorded on the "Classified Data Sheets",
example of which is shown in the above mentioned reports. The ratings for decay and
termite attack were determined in the same manner as explained in the previous reports.

In previous reports the average index rating was obtained by averaging the decay
rating with the termite attack rating. It has been pointed out that when all the stakes
have failed, either by decay or termite attack, the average index rating should be zero,
but the method previously used to determine this average rating will result in a num-
erical value greater than zero after all the stakes have failed. The method to determine
the average value in this report follows the recommendation of ASTM Method D 1758-
60T, Evaluating Wood Preservatives by Field Tests with Stakes, in which the decay
and termite grades are combined, that is, the lowest grade from either cause is used.
The average index rating for the untreated specimens as determined for the previous
inspections and shown in Tables 1, 2, and 3 have not been revised, however.

Results of Inspections

Untreated Controls

The decay and termite attack ratings with their average index ratings for the un-
treated controls installed in November 1957 are shown in Table 1 for exposures of 14
months, 25 months, 40 months and 52 months. The values shown in this table are the
average of 10 untreated specimens of each species, and it can be seen that after 52 months,
7 of the oak, 6 of the fir and all 10 of the pine specimens had failed. It can be seen from
the index ratings that both decay and termite attack contributed to the failure of the
specimens.

Additional untreated controls were installed at the time of each inspection, and the
index ratings for these specimens for the exposure time are shown on Tables 2, 3, and 4.
In general the index ratings for termite attack are lower than the ratings for decay.

Treated Specimens

The decay and termite attack ratings after 52 months of exposure for the speci-
mens treated with preservatives 1 to 9, inch, are shown in Table 5. The data are also
shown for preservative 13, a 25 percent creosote with 75 percent petroleum, after 27
months of exposure for the No. 1 retention. Complete data are shown for retention 1
and 2 for the nine preservatives and also for the specimens having No. 3 retention of
preservative 2, chromated zinc chloride, and preservative 3, tanalith. Only a cursory
examination was made of the remaining specimens, but it was evident that very little,
if any, decay or termite attack has taken place.

The data tabulated in Table 5 for the specimens treated with No. 1 retention are
shown graphically by the bar diagram on Fig. 1. It is evident from this diagram that
even with this light retention, which is only half that recommended by AREA, coal tar
creosote is offering the best protection against decay and termite attack while chromated
zinc chloride and tanalith are offering the least protection.

Conclusions

From the data secured during the inspection of treated specimens of oak, fir and pine
species after 52 months of exposure, it seems logical to conclude that:

Termite Control Investig ation

Table 1

SUMMARY OF INDEX RATINGS

UNTREATED SPECIMENS

INSTALLED: NOV. 1957

T=Termito

or
D» Decay

14 Months
Exposure ,

25 Months
Exposure

40 Months
Exposure

52 Months
Exposure

Oak

Fir

Pine

Oak

Fir

Pine

Oak

Fir

Pine

Oak

Fir

Pine

AV.

77.5
72.5

75.0

67.5
95.0
81.25

60.0
70.0
65.0

37.5
40.0

38.75

40.0
72.5
56.25

30.0
25.0
27.5

27.5
22.5
25.0

40.0
55.0

48.75

10.0
12.5
11.25

10.0

17.5

7.5

15.0
47.5

15.0

0
0
0

No. of
Failures

Table 2

SUMMARY OF INDEX RATINGS

UNTREATED SPECIMENS

INSTALLED: JAN 1959

T=Termite

or
D=Decay

11 Months
Exposure

26 Months
Exposure

37 Months
Exposure

Oak

Fir

Pine

Oak

Fir

Pine

Oak

Fir

Pine

AV.

62.5
67.5
65.0

60.0
67.5
63.75

62.5
65.0
63.75

47.5
42.5
45.0

37.5

57.5
47.5

37.5
17.5
27.5

30.0
35.0
25.0

30.0
42.5
20.0

5.0

No. of
Failures

Table 3

SUMMARY OF INDEX RATINGS

UNTREATED SPECIMENS

INSTALLED: NOV. 1959

T=Termite

or
D-Decay

15 Months
Exposure

27 Months
Exposure

Oak

Fir

Pine

Oak

Fir

Pine

AV.

70.0

75.0
72.5

67.5
77.5
72.5

55.0
70.0
62.5

57.5
65.0

50.0

45.0
67.5
45.0

35.0
47.5
20.0

No. of
Failures

Table h

SUMMARY OF INDEX RATINGS

UNTREATED SPECIMENS

INSTALLED: FEB. 1961

T"=Termite

D=Decay

12 Months
Exposure

Oak

Fir

Pine

AV.

62.5
87.5
60.0

50.0
75.0
50.0

47.5
72.5
40.0

No. of
Failures

1. Coal tar creosote is offering the best protection against decay and termite
attack.

2. Chromated zinc chloride and tanalith are affording less protection against
decay and termite attack than the other preservatives. However, even the
bwer retentions of chromated zinc chloride and tanalith considerably reduce
the rate of decay and termite attack compared with no treatment.

Termite Control Investigation

Table 5

SUMMARY OF QUALITY INDEX-RATINGS

TREATED SPE OMENS

Months

T=Termite

No.

1 Retention

2 Retention

No. 3 Retention

of
Service

Pres.

D=Decay

Oak

Fir

Pine

Oak

Fir

Pine

Oak

Fir

Pine

100.0

97.5

100.0

97.5

100.0

AV.

100.0

97.5

100.0

67.5

55.0

37.5

70.0

60.0

67.5

75.0

80.0

82.5

80.0

92.5

80.0

92.5

95.0

100.0

AV.

55.0

37.5

67.5

60.0

67.5

75.0

80.0

82.5

60.0

87.5

35.0

65.0

77.5

75.0

97.5

92.5

57.5

92.5

55.0

57.5

92.5

87.5

67.5

97.5

95.0

AV.

45.0

80.0

27.5

45.0

77.5

70.0

67.5

90.0

87.5

90.0

75.0

70.0

100.0

87.5

82.5

87.5

90.0

92.5

97.5

100.0

92.5

AV.

80.0

72.5

70.0

97.5

87.5

80.0

90.0

92.5

85.0

97.5

100.0

87.5

95.0

100.0

AV.

80.0

90.0

82.5

97.5

100.0

72.5

65.0

67.5

97.5

100.0

92.5

82.5

92.5

82.5

97.5

100.0

AV.

67.5

62.5

65.0

92.5

100.0

92.5

100.0

92.5

100.0

97.5

77.5

100.0

87.5

100.0

AV.

77.5

100.0

92.5

87.5

100.0

97.5

75.0

90.0

62.5

92.5

97.5

90.0

72.5

87.5

85.0

75.0

95.0

AV.

60.0

82.5

55.0

75.0

95.0

85.0

72.5

90.0

77.5

92.5

85.0

67.5

90.0

97.5

85.0

87.5

100.0

AV.

57.5

82.5

77.5

82.5

80.0

85.0

AV.

D
AV.

100.0

97.5

AV.

95.0
95.0

100.0
100.0

92.5
90.0

Specimens treated with preservatives 1 to 9 incl. were installed in November 1957.
Specimens treated with preservatives 10 to 13 incl. were installed in November 1959.

Inspection Date: Feb. 1962

Ter mite Control Investigation

FIG l

SUMMA 'ALITY INDEX RATINGS

TREATED SPECIMENS
RETENTION: I

QUALITf INDEX RATING- PERCENT
— DECAY
iTMBOl urn™ TERMITE
100 JO dO 70 60 50 40 iO

Advance Report of Committee 1 — Roadway and Ballast
F. N. Beighley, Chairman

Grading Problems Encountered During Relocation of the

Santa Fe Railway's Main Line Between Williams

and Crookton, Ariz.

By R. A. STANE
Construction Engineer, Coast Lines, Atchison, Topeka & Santa Fe Railway

The Santa Fe Railway completed the construction of a 44-mile relocation of its
double-track railroad between Williams and Crookton, Ariz., in December 1960. Con-
struction of the original single-track railrcad was started in April 1881, and the line
was completed and placed in service in the winter of 1883. Lewis Kingman, Santa Fe's
locating engineer at that time, was aware of a more operable route farther to the north.
However, its construction would have entailed the movement of malpais rock in quan-
tities too large to be practicable for the equipment of that time, so he was forced to
locate the line from Williams down through Johnson Canyon to Crookton by way of
Ash Fork. This location called for canyon wall construction and the building of Johnson
Canyon tunnel, 396 ft long and on a 10-deg, 6-min curve.

In time the single-track railrcad no longer sufficed, and by 1913 a separated tunnel-
free second track had been placed in service to serve as the eastbound main line as it
was intended that westward traffic would be carried on the old single-track railroad.
However, the company was frequently called upon to handle westbound loads too high
or too wide to pass through Johnson Canyon tunnel and they had to be run against
traffic on the eastbound main line, resulting in considerable loss of train time in eastward
movements. This, together with excessive costs due to rail wear on curves as sharp as
10 deg, 34 min, fuel consumption on heavy grades as steep as 1.8 percent, helper service,
etc., created a serious need for a major relocation.

Reconnaissance surveys were made, and it was determined that with modern equip-
ment a route employing curves of 1 deg or less and maximum grades of 1 percent was
feasible — probably not too distant from the northerly route mentioned in Mr. Kingman's
memoirs.

During the time of planning, the basalts were considered to be a major problem
even though they constituted only 41 percent of the solid-rock excavation, the remain-
ing 59 percent being sandstone and limestone. If the basalts had been entirely solid and
unbroken, they would not have been too difficult to handle with modern heavy equip-
ment and blasting agents. However, they were badly fractured during the cooling process,
and, in addition, many types were encountered, each presenting an individual problem.

The first type was solid and firm with no distinct cleavage or fracture planes. It
was so coarse in grain that it approached the texture of gabbro, and was dark in color.
On the surface, it was evidenced by well rounded outcrops. The rock was hard; the rate
of drill penetration in it was low, approximately 30 to 40 ft per hr, and the powder
factor was high — as high as 1.75 lb per cu yd. However, the rock broke into small
fragments, which resulted in good shovel production.

The second type was solid and firm with distinct, closely spaced horizontal cleavage
planes. It was fine grained and could be identified by its slab-like appearance and Bat
residual boulders. It could be drilled at a rate of 52 to 56 ft per hr and broke well into
small fragments.

26 Grading Problems— Santa Fe's Williams-Crookton Relocation

&&&&&.

Fig. 1— North face of Cedar Tank cut at Mile Post SQS1/^ This shows
conditions at the time it was decided to lay slopes back and do additional
benching because of a general failure of the cut slopes caused by petrostatic
pressures in this cut, which is 115 ft deep.

Grading Problems— Santa Fe's Williams-Crookton Relocation 27

The third type was solid and firm with fracture planes in all directions. It was fine
grained and could be identified by its irregular and well rounded outcrops. Drill pene-
tration was approximately 56 ft per hr, and the rock broke into small, well graded
fragments which made it particularly suitable for roadbed embankments.

The fourth type was vesicular but firm, having no distinct cleavage or fracture planes
The rock was light in both color and weight and drilled easily, but blasting results were
erratic and much secondary shooting was required.

The fifth type was solid and firm but intermixed with volcanic cinders. The loose
cinders caused the drill holes to plug during the drilling and hading operations, which
resulted in high drilling costs and poor loading patterns causing poor fragmentation and
requiring extensive secondary shooting. However, the resulting material, though expen-
sive in its production, made exceptionally good roadbed embankment.

The sixth type was a volcanic cinder with inclusions of malpais boulders or dikes.
Stripping was difficult, in some areas so much so that stripping was not economically
feasible. Drilling and loading were very difficult and expensive and recovery was unsatis-
factory. The resulting drill patterns were poor, and excessive secondary shooting was
required. This condition could be recognized by the presence of occasional ledges out-
cropping in the cinder areas.

Roadbeds were constructed in excavation with a minimum bottom width of 42 tt.
sufficiently wide to permit the use of 4-yd shovels. As these machines are capable of
working 30-ft faces, drill holes were, for the most part, 30 ft deep. The holes were 3in.
in diameter and were drilled on various patterns, the most common being on 6- by 7-tt
centers, and were loaded with ammonium nitrate blasting materials boosted by 60 percent
extra dynamite which was, in turn, detonated by primacord.

On the average, 1.347 lb of explosive were used per cubic yard of malpais excavated.
During the first V/2 years the track was in service the only failures experienced in
the malpais areas were in cut slopes comprised of malpais boulders in a matrix of vol-
canic cinders. These failures were minor and developed so slowly that loosened materials
could be removed before the track was endangered.

Sandstone and limestone comprised the greater portion of the rock excavation and.
though more easily moved than the malpais, they posed the most difficult design prob
lems. Both sedimentary formations were laid down during the Permian Period. The
sandstones are particularly interesting, as in them is seen substantiation of the theory
of shifting poles. According to recent papers by geologist M. D. Opdyke and physicisl
S. K. Runcorn a study of the magnetic fields in the Supai sandstone, southeast of Wil
liams and southwest of Flagstaff, indicates that at the time of their formation, during
the early Permian Period, the north pole was near Nanking, China, not taking into
account changes in relative location due to continental drift. This placed the equator
approximately in the middle of Utah. Its direction was that which is now northeast to
southwest. A later study of the magnetic fields, during the Triassic Period, in the Moen
kopi formation shows the north pole to have been so positioned as to have placed the
equator in the proximity of Tampico, Mexico in a direction which is now south IS deg
west, more or less. If this may be accepted, the equator passed slightly south ol the
Williams area about Mid-Permian in a direction roughly northeast to southwest, and
for this reason the prevailing winds wen- fr< m what is now north but at that time
northeast.

Sand migrations caused by these winds sweeping over an ancient dead sea built up
deposits which were later to become the Coconino sandstones. The materials so dej

28 Grading Problems— Santa Fe's Williams-Crookton Relocation

Fig. 2 — North side of cut about 2000 ft long and 75 ft deep just east
of Mile Post 410. This picture shows the failure which occurred in the
deepest part of the cut about one year after the line was placed in service.

Grading Problems — Santa Fe's Williams-Crookton Relocation 29

Fig. 3 — North side of same cut as shown in Fig. 2, but view shows the
north face of this cut just east of point of failure in the steeply dipping
sandstone.

created a huge sand plain, its surface being from 200 to 300 ft above underlying mate-
rials— in this area the Hermit shales — and, as equatorial winds are very nearly constant
in direction, a great sloping front was advanced toward the southwest as the sands
spilled over its face. This would account for the bedding planes dipping from what is
now north to south due to the 45 deg rotation of the earth's surface, with occasional
ancmalies caused by storms moving in directions contrary to that of the trade wind-
Laminations are probably the result of crustations caused by intermittent precipitation
and/or gradation variations resulting from wind velocity changes.

The general area then subsided and the Permian Sea moved in to form the Kaibab
limestone by marine depositicn. The country rose and again subsided for the Cretaceous
Sea. It then began its rise to its present level. This rise was fairly uniform over a great
area. However, minor differences caused alternate compression and tension in the rock
masses which probably accounts for the jointing now evident in plane- roughly °0 deg
to each other.

The sandstone lies in thin beds varying in thickness from less than 1 in to 2 or
3 ft. It is badly crossbedded and lies on planes sloping, for the mosl part, 2 or 3 tt
horizontally to 1 ft vertically. This variatii n from the u.-ual angle Ol repose of -and-
is probably due largely to the encouragement of horizontal movement oi the -and par
tides by winds passing over the sloping face always in a direction tending to Batten
the slopes. To a lesser degree, thi- Hatter slope is the result of the pull of gravity being
less as it would be at the equator because of the greater distance between the centers

Grading Problems — Santa Fc's Williams-Crookton Relocation

■

Fig. 4 — North face of cut just east of Mile Post 410 during the con-
struction period. The massive sandstone material jutting out of the slope
behind the pickup truck was removed by a series of tractor treads pulled
between two crawler tractors.

of masses, and greater opposition by an increased centrifugal force because of the ex-
tended radius of the earth at the equator and a slightly faster rotation of the earth at
that time, all of which would reduce the "weight" of the sand particles thus requiring
a lesser external horizontal force to cause movement.

The bedding planes are covered by a thin coating of clay which originated in the
overlying shales and limestones. The joints through which these clays gained access to
the bedding planes now permit surface water to find its way to the clays, thus lubricating
the planes making huge rock slides a possibility. In fact, approximately one year after
the new line was placed in service, a large-scale failure occurred following a period of
considerable precipitation. The failure, however, was anticipated at the time of construc-
tion and sufficient extra width had been provided to safely contain the slide.

In many areas the sandstone formation is still overlaid with limestone in beds
which lie very nearly on the horizontal. Apparently the entire area through which the
new line was constructed is presently in compression due to the current upward move-
ment of the earth's crust, as it is found that instability of slopes in limestone cuts was
further increased by petrostatic pressure. The blasting for one cut, approximately
40 ft deep and 3000 ft long, was done in one operation, and the limestone overlying the

Grading Problems — Santa Fe's Williams-Crookton Relocation 3 1

sandstone suddenly moved in about 14 in. No correction was made and no further
movement has occurred. Another cut, 13,000 ft long and having a maximum depth of
115 ft, was designed for a roadway width of 42 ft, s'.opes of JA:1, and a 25-ft bench at
the 50-ft level. Excavation was about two-thirds completed when the slopes began to
fail due to petrostatic pressures, and it was necessary to reopen the entire cut. In the
redesigned roadbed the original width and slopes were maintained; however, two 35-ft
benches were constructed at the 40-ft and 70-ft levels. The redesigned slopes, during
the first two years after construction, have been stationary, and it appears that there
will be no further movement.

The sandstones, though highly abrasive to bits, could be drilled easily, progress
being from 62 to 70 ft per hr.

Limestones were not abrasive and varied considerably in hardness. The harder
limestones were drilled at the rate of 54 to 60 ft per hr; the softer limestones were
drilled at rates varying from 80 to 100 ft per hr.

On the average, 1.075 lb of explosive were used per cubic yard of sandstone/lime-
stone excavated.

Over the entire job the blasting agents were comprised, by weight, of 13.5 percent
dynamite and 86.5 percent ammonium nitrate in one form or another. Coated ammonium
nitrate prills were used where holes were dry; however, nitro-carbo-nitrate was used
where holes were wet. It was found, when the holes were dry, that the nitro-carbo-
nitrate materials were a little more efficient than the ammonium nitrate prills sensitized
by the addition of 2 qt of diesel fuel oil per 50 lb of prills. Regardless as to which
materials were used, the required spacing of dynamite bocs.ers was the same — in gen-
eral, three 1.09-lb sticks in the bottom of the hole with 1 stick at every 5- to 8-ft level
to within 30 in of the top of the hole.

A definite saving can be made by making use of ammonium nitrate prills wherever
possible. For example, letting the cost of coated ammonium nitrate prills after treatment
serve as an index, that is 1.0, the comparative cost of one of the nitro-carbo-nitrates
was 1.57, the second 2.22, and dynamite 4.83.

Usual railroad roadbed specifications require that embankments made of rock be
constructed in lifts of no greater than 18-in depths. To have obtained rock usable in
such lifts would have required much more expensive drilling and blasting. Dense fill-
were constructed using rock of 3-ft sizes and even, occasionally, as large as 4 ft in
diameter. This was accomplished by dropping the loads on top of the lift being made
and dozing the materials over the face of the lift. The fines of the later dumpings were
thus dropped into the voids created by the larger rolling rock of the preceding dumpings
Occasionally it was necessary to bring in fines to provide a smooth roadwaj for the
trucks, and only when this was done were sheepsfoot tamp- employed on rock fills.

No water was used on the entire construction, including the making of roadbed
embankments of cinders and clay, and, by slightly increasing the compactive effort, no
difficulty was experienced in obtaining required densities. For the most part, the com
paction of common materials was by proper routing of beavj rubber-tired hauling
equipment, the remainder being by sheepsfoot tamp.

The compaction of select materials around corrugated metal pipes was also accom
plished without the use of water. These materials, either cinders deteriorating to daj
or of a more solid granular nature, were tamped with hand-operated pneumati» tamper-
directly under the haunch of the pipe. The volume so tamped was kept at an absolute
minimum and the remainder, up to the spring line, was compacted bj a No. 12 motor

32 Grading Problems— Santa Fe's Williams-Crookton Relocation

Fig. 5— North face of 75-ft cut just east of Mile Post 410 during the
construction period, showing the variable dips and strikes of the Coconino
sandstone of the Permian period.

Grading Problems — Santa Fe's Williams-Crookton Relocation 33

^^SvSSSS^

► *..-» -~4

M?0

*k-

i>isf:

Fig. 6 — North face of sandstone cut with maximum depth of 110 ft
just west of Mile Post 410, showing condition of the Coconino sandstone
during the construction period.

patrol running longitudinally with the pipe with its front wheels laid over as far as
possible so as to force the material under the pipe. Above the spring line, to an elevation
of twice the radius above the spring line but not less than 3 ft over the top of the
pipe, and outside of vertical planes tangent to the pipe, materials were tamped by a
front-end loader running longitudinally with the pipe and carrying a full load in its
scoop. In so doing, loose materials spilled over the top of the pipe, filling that area
between the vertical planes mentioned above. The leader was then passed transversely
over the pipe until at least the upper 1 ft had been compacted to roadbed embankment
requirements. The result was an evenly compacted embankment easily meeting the re-
quirements for roadbed construction and the pipes were squeezed so as to slightly elon-
gate the vertical axis. It was found after large diameter pipes had been subjected to loads
imposed by high fills that the entire ellipse provided in the manufacture of the pipe
remained.

Since the line has been placed in service a slight Battening of the upper half of 72-in
corrugated metal pipes has been noted where they have been placed with minimum
— 3 ft below base of rail. This has occurred wherever select cinder material contained
an appreciable clay fraction. Apparently the clay fraction reduces the safety factor to
very nearly 1.0 percent. Maximum distortion probably occurs just before the front
wheels of fast-moving locomotives reach the center of the pipe, thus causing maximum

Hull. 573

34 Grading Problems — Santa Fe's Williams-Crookton Relocation

loading on one side of the pipe with nothing but the weight of the minimum cover
resisting on the other side. As all 72-in pipes under minimum cover did not fail it would
seem that the little additional resistance required may be obtained by increasing the
minimum cover approximately 6 in wherever the clay fraction cannot be avoided. An-
other solution to this problem might be the introduction of vibratory equipment above
the spring line to attain a higher degree of compaction directly over the pipe, as con-
siderable consolidation would be obtained through the transmission of vibratory forces
into the soils not directly beneath the compactor.

The Williams-Crookton relocation required the movement of 3% million yards of
basalt, 5 million yards of sandstone and limestone, Zl/2 million yards of common mate-
rials, \l/2 million yards of common borrow, the construction of bridges costing $2 mil-
lion, the laying of 88 track miles of electrically welded rail, and the installation of the
most modern train control and signal systems, certainly a major undertaking for private
enterprise and a clear indication that railroads are willing to expand their plants even
at tremendous costs to better their service in the field of heavy, long-haul transportation
for which they are so well equipped.

Advance Report of Committee 1 — Roadway and Ballast
Report on Assignment 10 (a)

Ballast Tests

T. W. Creighton (chairman, subcommittee), E. W. Bauman, J. G. Campbell, J. E. Gray,
\V. C. McCormick, E. W. McCuskey, E. L. Robinson, Jr., Stanton Walker, C. E.
Webb, E. L. Woods.

Your committee presents as information a report on conductivity tests conducted
at the AAR Research Center on open-hearth slag* furnished by the United States Steel
Corporation from Lorain, Ohio, to determine if that material is satisfactory for use as
railrcad ballast in track-circuit territory.

This research is sponsored by Committee 1 — Roadway and Ballast, and the tests
were performed by the engineering research staff of the AAR Research Depart nun t
under the direction of G. M. Magee, director of engineering research, and Rockwell
Smith, research engineer roadway. The tests were conceived and conducted by M. F.
Smucker, assistant electrical engineer, and G. L. Hinueber, engineering laboratory man-
ager. The report was prepared by G. L. Hinueber.

A series of conductivity tests were run on the open-hearth slag using a specially
built 8-cu-ft box with dimensions of 2 by 2 by 2 ft. The box was constructed of >4-in ply-
wood, and the inside was coated with epoxy resin to waterproof it. Two copper plates
(24 in by llj^ in by iV in) were bonded to each of two opposite faces of the inside
of the box. The upper and lower plates on each face were separated by approximately
1 in.

The box was filled to the top with the open-hearth slag and the resistance between
the copper plates on opposite faces was determined by using an ohmmeter. Resistance
readings were taken between the two bottom plates and between the two top plates.
The top and bottom plates were then connected in parallel, and resistance readings were
then taken between the two sides. The open-hearth slag was removed from the box and
sprinkled with water until thoroughly wet. It was allowed to drain for about 1 hr and
then shoveled back into the box. The resistance of the wet ballast in the box was deter-
mined by applying a-c voltage across the copper plates, measuring both the current and
applied voltage. The dynamic resistance of the sample was determined by dividing the
applied voltage by the current. It was necessary to use the a-c method for determining
the resistance of the wet ballast material to overcome the polarity of the individual
ballast particles which would tend to make d-c resistance readings unreliable.

In order to compare the resistance readings of the opm-hearth-slag ballast material
with a ballast of acceptable conductivity, the procedure described above was repeated,
using a representative blast-furnace-slag ballast.

The results of these tests are as follow-:

Ballast Type
Open-hearth slag
Open-hearth slag
Open-hearth slag
Blast-furnace slag
Blast-furnace slag
Blast-furnace slag

* Open-hearth slag is formed simultaneously whin steel i- produced and refined i" open-hearth
furnaces. It consists essentially of a fused mixturi of oxides and

Plata Between

.' Readings

Whirh Resistance

Box Full

Full

Was Measured

Empty Box

Dry Ballast

Wei Ballast

. Bottom in bottom

Infinity

620 megohms

3500 ohms

.Tup In tOP

infinity

620 megohms

3900 ohm-

.Side to side

infinity

500 megohms

2200 ohm-

.Bottom to bottom

infinity

I n G n i t y

1850 ohm-

.Top to top

infinity

Infinity

3100 ohms

.Side to side

Infinity

1350 ohm-

36 Ballast Tests

These results indicate that the open-hearth slag is a better conductor in the dry
state than the blast-furnace slag. This was to be expected because of the higher metallic
content of the open-hearth slag. However, when the ballast materials were wet the
open-hearth slag showed a slightly lower conductivity than the blast-furnace slag. Blast-
furnace slag is more porous than the open hearth and consequently will retain more
water, which is a conductor. It will also be noted that the b'.ast-furnace slag sample
contained more fine material than the open-hearth slag, and a check with a magnet indi-
cated the presence of considerable metallic material in the fines. These fines had a tend-
ency to settle to the bottom of the test bcx and probably account for the resistance
reading of the wet blast-furnace slag in the bottom half of the box being lower than
that in the top half.

The next phase of the investigation consisted of testing the conductivity of individual
pieces of seven different ballast materials. Ten representative pieces of each of the seven
ballast materials were selected. An chmmeter was used to determine the resistance across
each piece in the dry condition. Next the pieces were soaked in water and the resistance
across each piece was determined for the wet state. The pieces were then surface dried
with a cloth and resistances determined for the saturated surface-dry state. The fol-
lowing results of these tests are reported for the average of the resistances of the 10
pieces of each ballast material:

Saturated Surface- Wet

Ballast Material Dry Resistance Dry Resistance Resistance

Open-hearth slag 27 megohms 550,000 ohms 145,000 ohms

Blast-furnace slag Infinity 850,000 ohms 180,000 ohms

Limestone Infinity 1.4 meg ohms 260,000 ohms

Chert (Chat) Infinity 10 meg ohms 500,000 ohms

Quartzite Infinity 10 meg ohms 800,000 ohms

Trap Rock Infinity 10 meg ohms 600,000 ohms

Granite Infinity 10 meg ohms 340,000 ohms

It will be noted that a considerable difference exists between the resistance of the
open-hearth slag pieces and the resistance of the other types of ballast tested in the dry
state. This difference is somewhat less, although still considerable, in the saturated surface-
dry state. The difference between the resistance of the open-hearth slag and the resistance
of the other ballast materials tested is considerably smaller in the saturated state. The
materials with higher porosities, such as blast-furnace slag, and higher absorptive qual-
ities, such as limestone, exhibit conductivity of higher magnitudes in the saturated surface-
dry and wet states than those of lower porosity and lower absorption. The conductivity
of these more porous and absorptive aggregates approaches the conductivity of the
open-hearth slag fairly closely for the saturated surface-dry condition and even more
closely for the wet or saturated state.

The last phase of the investigation was conducted with the use of a 5-ft, 3-tie sec-
tion of railroad track. The track was first placed on the laboratory floor and the re-
sistance across the rails was read with an ohmmeter. The dry open-hearth slag ballast
was then leveled to a depth of 2 ft and the track section placed on top. The resistance
reading across the rails was again read. Next, the cribs between the ties were fiiled with
ballast, making certain that the ballast particles were in contact with the rail base. The
resistance reading was repeated. The ballast and the track section were then thoroughly
saturated with water and the resistance reading repeated once again. The track and
ballast section were allowed to drain for a period of about 1 hr, at which time the
resistance reading was taken for the damp condition.

Ballast Tests 37

Results of these tests on the open-hearth slag are as follows:

Resistance
Condition Across Rails

3-tie section on lab. floor 20 megohms

3-tie section on top of dry ballast 10 megohms

J dry ballast 10 megohms

3-tie section on ballast with cribs full I damp ballast 7500 ohms

j wet ballast 4500 ohms

Again it will be observed that the lowest resistance measured was for the thoroughly
saturated condition. If we extrapolate the resistance for the 5-ft track length to a 1000-ft
track length, we get a 22. 5-ohm resistance for the saturated condition. This value satisfies
the minimum value of 2 ohms as specified by the AAR Communication and Signal
Section. The tests utilizing the 5-ft track section were, however, run on the laboratory
flcor, and it should be noted that the test procedure is of limited extent and would not
necessarily be directly comparable to track conditions.

In 1959 three ballast materials, a limestone, an open-hearth slag from Hammond.
Ind., and a blast-furnace slag, were tested for conductivity using the 5-ft track section
and following the same procedure as described above. The resistances for the saturated
condition per 1000 ft track length were obtained by extrapolating the data from the
5 ft track length. The results were as follows:

Resistance —

Ohms
Per 1000 Ft
Ballast Material Track Length

Limestone (saturated) 22.5 ohms

Open-hearth slag (saturated) 15.0 ohms

Blast-furnace slag (saturated) 20.0 ohms

All of the tests run indicate that the open-hearth slag sample had a higher conduc-
tivity in the dry state than any of the other ballast materials tested. However, in tin
wet condition, which is the most critical as far as conductivity is concerned, the open-
hearth slag is about comparable to the blast-furnace slag used in the investigation and
apparently not much less favorable than the limestone. In fact, the resistance of the
individual pieces of the open-hearth slag tested was in the same general range as. although
lower than, the quartzite, chat, trap rock and »ranitc matt-rials whin tested in tin-
saturated condition.

The test results indicate that the moisture in the ballast cm In- a more important
factor in conductivity than the ballast material itself. For the- ballast materials tested,
the moisture is undoubtedly the governing factor when the ballast section is saturated.

In addition to the conductivity tests, the tests required in the AREA Ballast
Specifications for air-cooled blast-furnace slag were run en the open-hearth slag from
Lorain, Ohio. These test results are as follows:

( in \n\l [ON

Screen Size Percent Passing

2]/2" 100.0

V/2" 80.9

y4" i6A

V2" 2.6

H" 1-7

No. 4 1.2

38 Ballast Tests

Los Angeles Abrasion

Percent loss =31.5

Sodium Sulfate Soundness

Percent loss= 1.0

Weight per Cubic Foot (Compacted)

Wt/Ft3= 116.6 lb

The gradation does not conform strictly with any of the AREA ballast grading
requirements. However, this is merely a matter of processing which could be easily
altered to produce a material of satisfactory gradation. All of the other physical test
results are well within the limits required by the AREA Ballast Specifications.

The test results on the open-hearth slag as included in this report are for a sample
furnished by the U. S. Steel Corporation from Lorain, Ohio. Although these results indi-
cate that this material should prove satisfactory from the conductivity standpoint, it
must be noted that they will not necessarily apply to open-hearth slag material from
other sources. It is believed that open-hearth slag can be processed to produce acceptable
ballast material. However, control of the processing, particularly for the elimination of
particles high in metallic content, is very important.

In addition to the above tests a field inspection was made on a railroad in Michigan
whose main line track is ballasted for a distance of approximately 90 miles with an open-
hearth slag ballast. The track has been in service with this ballast for a period of about
five years without any difficulty being encountered with the signal system. The ballast
is performing well in service, having apparently resisted breakdown due to weathering
and abrasion. Inspections of this nature are planned on other railroads using open-hearth
slag ballast, with samples being taken for laboratory tests.

Rail Slippage Tests — Concrete Ties*

Introduction

In designing the fastenings for AAR Type E prestressed concrete ties, and having
in mind particularly their use with welded rail, it was considered that a minimum rail
slippage resistance of 2000 lb per tie per rail was desirable. This was based on measure-
ments that had been made of the resistance of wood ties to movement in the ballast by
rail creepage forces (see AREA Proceedings, Vol. 56, 1955, page 283) and on studies
made of the forces needed to properly restrain welded rail from expansion and contrac-
tion movements at the ends, against buckling in extreme hot weather, and against exces-
sive gap in the event of rail breakage in very cold weather (see AREA Proceedings,
Vol. 38, 1937, page 493). Calculations based on a coefficient of friction of the rail base
on the tie plate of 0.25 and the leverage ratios of the clips indicated that with the
fastenings designed for the Type E tie the slippage resistance would be 2000 lb with
5000-lb bolt tension. It was further considered that an initial bolt tension of 10,000 lb
would be desirable to extend the period between tightenings so the cost of tightening
the clip bolts would be within economical limits. It was, therefore, anticipated that the
slippage resistance would range between 4000 and 2000 lb per tie per rail.

However, no actual slippage tests had been made with the fastenings to check the
slippage resistance anticipated in the design calculations. For making these tests it was
thought desirable to include a drive-on type anchor and wood tie, and also French,
German and Swedish concrete ties with the particular type fastenings used on these ties.
In discussion of the subject, the Portland Cement Association kindly agreed to the con-
duct of such tests in their research development laboratories at Skokie, 111., where the
PCA staff designed and built the test layout, and provided the necessary test instru-
mentation and a portion of the personnel for operation. The AAR furnished the ties
and fastenings, except the Swedish tie, and members of the research staff cooperated
in the conduct of the tests.

Test Procedure

It was decided that the control test should be with conventional wood tie construc-
tion using 132 RE rail, a double-shoulder tie plate and a grip-type rail anchor. The
control test was followed by the AAR Type E prestressed concrete tie with AAR design
fastenings, a variation of the Type E tie without a tie plate, but having direct fixation
with only a pad under the rail, the French RS concrete tie, the German post-tensioned
concrete tie with GEO-type fastenings, and the Swedish concrete tie with FIST sprint
rail clip. The fastenings for the concrete ties are shown in Fig. 1.

It was believed that the vibrations that accompany train movement might have an
effect on the rail slippage resistance of the fastenings. In order to study the vibration
effect, certain of the tests were repeated with a 60-cycle vibrator attached to the rail;
the magnitude of vibration was determined by the amplitude of vibration <>t the rail
directly above the tie connection. This method of introducing vibration was used for
practical reasons realizing that the vibrations so produced did not necessarilj simulate
those that occur in track.

The testing layout was designed for tests as follows with three slips for each con

dition.

*An abstract of Report No. V.R-22 issued by the Research Department. \ \k. Copies of the full
report can be obtained from the director of engineering research, AAR, 3140 a Chicago 16.

Rail Slippage Tests — Concrete Ties

AAR Tie - AAR Fastenings

AAR Tie - Direct Fixation

French RS Tie & Fastenings

German Tie - GEO Fastenings Swedish Tie - FIST Clip

Fig. 1.

Rail Slippage Tests — Concrete Ties 41

Series A — Wood tie control test

2 rail anchors without vibration

2 rail anchors with vibration

Total 4 tests
Series B — Concrete tie, Type E, AAR fastenings (See ER-20 report for details)

Bolt tension 5000 lb, 7500 lb, 10,000 lb

Polyethylene pad without vibration

Polyethylene pad with vibration

Hardwood plywood pad without vibration

Total 9 tests
Series C — Concrete tie, variation of Type E for direct fixation with uniclips

Bolt tension 5000 lb, 7500 lb and 10,000 lb

Polyethylene pad without vibration

Polyethylene pad with vibration

Hardwood plywood pad without vibration

Total 9 tests
Series D — Concrete tie, French RS with rubber pad and RX clips

With and without vibration

2 tests
Series E — Concrete tie, German with GEO-type fastenings

With and without vibration

2 tests
Series F — Concrete tie, Swedish 101 with rubber pad and FIST rail clip

With and without vibration

2 tests

The test layout, shown in the illustration on page 42, consisted of a concrete
base with a variable position loading crosshead for attaching a hydraulic jack with pro-
vision for centering the jack at the centroidal axis of the rail. Hold-down bolts, suitably-
spaced in the concrete base, permitted moving the jack and rail over a sufficient range
to obtain new bearing surfaces on the rail and tie plate or pad for each test. The tie was
sjrouted to the floor at proper height and held down by two crossheads bolted through
the floor by tensioned tie-rods which were prestressed by hydraulic jacks beneath the
floor. Thus the slippage values are between the rail and the tie or between the rail and
the tie plate without movement of the tie. The free end of the rail was supported on a
ring of roller bearings to reduce friction to the minimum. Pressure on the rail end was
applied at 2000 lb per minute.

The instrumentation consisted of a load cell to measure the force at the rail end
while the movements of the rail with respect to the tie and also with respect to the tie
plate were determined by differential transformers. AH instrumentation results were
continuously recorded by an oscillograph. Dial gages were used with the grip-type r;iil
anchor to determine the incident of slip of the rail through the anchor.

Vibration was provided by a 60-cycle vibrator mounted <>" the rail. The rail ampli-
tudes above the tie were those which could be held constant with the equipment

The actual bolt tension was measured by means of two -train gages mounted 00
each bolt throuuh the use of a static strain Indicator, In the case of the French tie,
the bolts were tightened to a prescribed torque of 120 ft-lb and the tension was
measured with the strain gages, except for slip test D-4 where the bolts were tightened
to a tension of 7000 lb.

Rail Slippage Tests — Concrete Ties

Test layout.

Rail anchors, clips, plates and pads were changed out after each test to provide new
bearing surfaces.

The AAR is indebted to the Portland Cement Association for providing the test
facilities and the personnel to aid in carrying out the investigation. The tests were con-
ducted in the PCA laboratory under the general direction of Eivind Hognestad, manager.
Structural Development Section. The design and construction of the test equipment was
by Charles H. Raths, assistant development engineer, who also participated in the direct
supervision of the tests.

Participation in the tests by the AAR research staff was under the general direction
of W. M. Keller, vice president— research, and G. M. Magee, director of engineering
research; E. J. Ruble, executive research engineer, was in direct charge, assisted by L. R.
Lamport, assistant research engineer — track, who shared the supervision at the PCA
laboratory and was responsible for the report.

Test Results

The slippage forces together with the initial bolt tension and bolt tension loss for
each test condition are presented in Report No. ER-22 in Tables 1, 2, 3, and 4, which
are not reproduced herein. In the test with the Swedish tie an attempt was made to
measure the clamping force on the rail by means of strain gages but the results were
unsatisfactory, hence this report must deal only with the slippage force applied to the
rail end. The Swedish Railway, however, advises that the pressure on the rail is 2000 kg
(4410 lb).

Rail Slippage Tests — Concrete Ties

RAIL SLIPPAGE TESTS
Table 5. Comparison of Average Slippage Forces

Type of Tie
and Anchor or Clip

Witho

it Vibration

With

Vibration

Slip Force

Percent of Control

Slip Force

Percent of Control

Creosoted Oak Tie- Fair

Rail Anchor Control

5215

100

5445

100

AAR Type "E" Concrete

Tie, AAR Clips

Polvethvlene Pad

5000 lb. Bolt Tension

2033

1002

7500 lb. Bolt Tension

3207

1717

10000 lb. Bolt Tension

4037

2707

Hdwd. Plywood Pad

5000 lb. Bolt Tension

2277

7500 lb. Bolt Tension

2217

10000 lb. Bolt Tension

3290

AAR Type "E" Concrete

Tie with Direct Fixation

and Uniclips

Polyethylene Pad

5000 lb. Bolt Tension

1197

1273

7500 lb. Bolt Tension

2163

1227

10000 lb. Bolt Tension

2593

2443

Hdwd. Plywood Pad

5000 lb. Bolt Tension

2013

10000 lb. Bolt Tension

3427

French Concrete Tie and

Clips (120 ft. lb. Torque)

2903

2320

German Concrete Tie and

GEO Clips

4143

3967

Swedish Concrete Tie and

FIST Clip

2160

2113

From the comparison in Table 5 it is apparent that rail slippage resistance with the
concrete tie is greatest on a steel tie plate, namely, the German tie with GEO fastening!
and the AAR tie with AAR design fastenings. These arc followed bj the AAR tie with
direct fixation on a hardwood plywood pad, the French tie with a rubber pad, the AAR
tie with direct fixation on a polyethylene pad, and last, the Swedish tie with a
rubber pad.

The loss of slippage resistance due to vibration was greatest "ii the AAR tie-, p.n
ticularly those with 5000-lb and 7500-lb bolt tension. There was no appreciable effed
on the rail anchors with wood ties, the German ties, or the Swedish ties.

The loss of bolt tension was somewhat erratic on the AAR Type E ties with the
AAR clips at 5000-lb and 7500-lb bolt tension. It is believed that this was due primarily
to some rotation of the clips as the slips occurred. More uniformity WW had in tin t v - 1 ~

Rail Slippage Tests — Concrete Ties

RAIL SLIPPAGE TESTS - CONCRETE TIES
Table 6. Ratio: Slippage Force to Pressure on Rail - Without Vibration

Tie and Fastenings

Initial Bolt
Tens ion -Lb.

Avg. Slip
Force

Avg. Pressure
on Rail

Ratio Slip Force
to Rail Pressure

AAR Type "E" with AAR
clips and polyethylene
pad

5000

7500

10000

2033
3207
4037

3770
5470
7810

0.54

0.59

0.52

Avg. 0.55

As above with hard-
wood plywood pad

5000

7500

10000

2277
2217
3290

3850
5810
7950

0.59

0.35

0.41

Avg. 0.45

AAR Type "E" direct
fixation with uni clips
and polyethylene pad

5000

7500

10000

1197
2163
2593

3680
5430
7220

0.33

0.38

0.36

Avg. 0.36

As above with hard-
wood plywood pad

5000
10000

2013
3427

3680
7220

0.55

0.47

Avg. 0.51

French Tie and fasten-
ings-rubber pad

120 ft. lb.
torque

2903

4760

0.61

German Tie-GEO
fastenings

6720

4143

5870

0.69

Swedish Tie- FIST
Clip

2160

4410

0.47

Vertical Pressure on Rail

AAR Type "E" with AAR Clips 2 x 0.41 x Bolt Tension
AAR Type "E" direct fixation-Uniclips 2 x 0.37 x Bolt Tension
French 2 x 0 . 44 x Bolt Tension
German 2 x 0. 50 x Bolt Tension
Swedish * 44i0 lb.

* Swedish Railway advise the pressure on rail is 2000 Kg (4410 lb. )

made with 10,000-lb bolt tension and also with the AAR Type E ties with direct fixa-
tion and the French and German ties, where little rotation of the clips was possible.
It will be noted that the bolts were retightened in a few instances with the result that
the bolt tension loss was usually less in the subsequent slip.

Table 6 shows the ratio of slippage force to pressure on the rail and brings out the
relatively low slippage resistance of the polyethylene pad on the AAR Type E tie with
direct fixation. The polyethylene pads have a smooth, wax-like surface which undoubtedly
accounts for the low slippage resistance.

Rail Slippage Tests — Concrete Ties 45

Discussion of Test Results

The values of rail slippage resistance as determined in these tests for the AAR design
clips for the Type E tie were generally in excess of the 2000 lb minimum for 5000-lb
bolt tension, except those with vibration, which were only about 1C00 lb. However, as
previously stated, it is not known to what extent the vibration actually simulated track
vibrations so it is questionable as to how significant these values are.

The slippage resistance with the AAR clip is much below that of the drive-on-
type anchor. This may be reflected in track by some slippage through the clips at the
ends of welded rail at extreme temperatures, but not necessarily to the extent the slip-
page resistance provided may be sufficient to move the ties in the ballast. The slippage
resistance provided throughout the length of the rail should be adequate to prevent
buckling and excessive gap opening in the event of rail breakage.

The slippage resistance with the direct fixation Uniclips is somewhat lower than for
the AAR design clips, and it might be found desirable to maintain the bolt tension
between a range of 7500 lb to 10,000 lb to provide adequate slippage resistance in track.

The slippage resistance with the German tie and GEO type clip was very good and
was little affected by vibration.

The slippage resistance for the French RN-type clip was slightly above that for
the AAR clip at the same bolt tension, possibly due to the two contact points between
the clip and rail and the shear force carried by the pad.

The slippage resistance of the Swedish clip was above the 2000-lb desired minimum,
both with and without vibration.

Conclusions

The measurements of rail slippage resistance obtained in these tests agree reasonably
well with those anticipated in the design of the fastening for the AAR Type E pre-
stressed concrete tie. It is believed that a range of bolt tension between 10,000 and
5000 lb should be maintained with AAR clips until such time as service experience might
indicate that a change in these limits is warranted.

Preventing Rail Failures in Track*

By G. M. MAGEE
Director of Engineering Research, AAR

Rail is the most important component of the track structure for the safe operation
of trains. It serves two important functions, one to support the wheel and the other
to guide it. In order to support the wheel, the rail, through its flexural strength and
stiffness as a beam, distributes the wheel load over several ties lengthwise of the track.
The ties in turn distribute the portion of the wheel load they receive to the ballast and
the ballast distributes this load to the subgrade. Thus the rail, ties and ballast serve to
distribute the wheel load over a sufficient area of the subgrade that the pressure per
square foot is well within its supporting power. However, in serving this function there
is a repeated flexural stress in the rail for each passing wheel load.

In order to serve its function of guiding the wheel, it is most important that the
rail provide a continuous path for the wheel. This requires that the rail shall not become
broken. If the rail becomes broken, the rail ends will not be held at the same height.
If the rail is under tension due to low temperature it will tend to open up and make
a gap when the rail breaks. Thus a broken rail may result in a derailment of a train,
although this is not necessarily so as there are many broken rails that occur in track
that do not result in derailments, particularly it the break occurs on tangent track or on
the low side of a curve. Nevertheless it is a matter of first importance in the safe opera-
tion of trains to prevent broken rails occurring in track.

How can this be done? Very few of the broken rails that occur in service are sudden
or complete fractures of the rail unless there is some defect present in the rail. Com-
plete and sudden fractures rarely occur in sound rail. Also, in most cases defects in rail
are originally small and progressively grow under repeated wheel loadings. Thus if we
are able to examine the rail carefully and periodically to determine whether there are
any flaws or defects in the rail of a size likely to initiate a sudden rupture and remove
such rails from track and replace them with sound rails, the occurrence of broken rails
can be largely prevented. To distinguish between rails that are removed from track
because of detected defects that may cause the rail to break and those that actually
break in track, we refer to the first as detected rail failures and to the latter as service
rail failures.

For the purpose of describing the development of equipment For detecting rail de-
fects, I should like to classify rail defects into two general types, namely, head d
and web defects. Head defects include transverse fissures, compound fissures, detail
tures, vertical split heads, and horizontal split heads. Although each of these types <>i
defects have certain distinguishing characteristics, they are all progressive fractures that
grow from a small nucleus and result in a separation of metal in the rail head. Also,
it is usually not possible to determine the existence of these defects by looking at tin-
rail because the separation of the metal is inside the rail head and therefore not visible

Web defects generallj consisl of cracks in the horizontal plane in the upper fillet
between the rail head and the web or in crack- out of boll holes Most defects ol this
type occur within joint bar limits Horizontal crark- in the upper web fillet are called
head and web separations. These sometimes develop on the gage ride of the low rail
of curves when the rail head has become considerably reduced in depth by wear and

•Presented before the Railroad Sessions of the Transportation Engineering < I the

American Society of Civil Engineers in Detroit, Mich., October 8-11, 1962. 1!

features in the program of the Railroad Sessions, on October 9. developed by the Board of Direction

of the AREA.

Preventing Rail Failures in Tr ack

The first rail defect detector car— 1928.

flow. These develop relatively slowly and can be detected readily by visual inspection
so they do not offer much of a problem. The head and web separations and bolt hole
cracks within the rail joint limits, however, are not readily accessible for inspection.
The cost of removing joint bars to make inspections periodically would be prohibitive.
Thus an adequate and economical means of inspecting the rail web within the joint
bar limits is needed for this purpose. Here again both of these types of defects are
progressive-type fractures that grow with repeated stressing from the passing wheel loads.

Defect-detection equipment was developed first for the head type of defect because
it was of the most importance. The first rail defect detector car was built by the Sperry
Development Company for the American Railway Association upon the recommenda-
tion of the Rail committee of the American Railway Engineering Association. Several
years were required for the research and development work, and the car was accepted
by the Committee on October 2, 1928. In this detector car, two sets of brushes for each
rail were mounted underneath to continuously contact the top of the rail as the car
moved along the track. A very heavy electrical current (2 volts, 2000 amperes) was
passed between the brushes through the rail. If a defect existed in the rail head, it
caused an abrupt change in the direction of the flow of current at the location of the
defect. A specially designed pick-up coil was located above the rail between the two
brushes. The abrupt change in the direction of current flow at a defect and resultant
change in the magnetic flux surrounding the rail head, generated a small potential in
the pick-up coil by inductance. The voltage from the pick-up coil was, of course, very
low and had to be amplified with radio-type amplifiers to provide enough power to

Preventing Rail Failures in Track 49

actuate recording pens on a moving paper tape. A paint gun was also connected with
the defect indicating mechanism so that when the pens were actuated a spot of yellow
paint was also sprayed on the rail to locate the defect. This principle of detection is
generally referred to as the inductance method.

This principle of defect detection has both advantages and disadvantages. The prin-
cipal advantage is that it gives a strong signal for a defect and thus can be used to
detect relatively small transverse defects in the rail head. One disadvantage is that an
electrical contact must be made between the brushes and the rail and sometimes the
surface condition of the rail makes this difficult. Also, if sparking should occur this
creates a localized magnetic effect that may give a false indication of a defect. Another
disadvantage is that when one brush has passed over the rail gap the defect detection
equipment will not be operative until the second brush has also passed over it, thus
leaving a portion of each rail end in which defects cannot be located.

Another type of the defect-detection equipment for locating defects in the rail head
was later developed by the Association of American Railroads and is known as the
residual magnetic method. With this method no electric current is actually passed through
the rail, but a powerful magnet placed toward the front end of the detector car on each
side passes a heavy magnetic flux through the rail head. After the magnet has passed,
if there is any defect in the rail head there will be a small residual magnetic field at
the defect. A pick-up coil mechanism located at the rear end of the detector car on each
side generates a small potential when it cuts through the lines of the residual magnetic
force. This signal is amplified to actuate pens on a recording paper tape and also to
spray a spot of paint on the rail. The advantages of the residual magnetic method are
that no current is actually passed through the rail so there is no problem offered by any
surface condition of the rail that would tend to affect the resistance of the brush con-
tacts. Also with the residual magnetic equipment it is possible to locate defects close to
the rail ends. In other words, there is very little dead area at the rail ends with the
AAR type car.

The first of the AAR type detector cars was a double-unit rail car, the first car
being the tow car and power supply and the second or trailer car having the defect
detection pick-up equipment and sleeping quarters for the crew. A recent development
of the AAR type equipment is the road-rail unit which utilizes the same general prin-
ciples of defect detection as the rail-type unit. However, every effort has been expended
to make the equipment compact and light in weight so that it can be applied to a truck
chassis with a specially built panel bcdy. This unit is equipped with retractable pilot
wheels so that it can be operated on the rail for testing or on the highway for trans
ferring between test location.

Today, rail in main track is tested periodically by detector cars, the frequency of
testing depending upon the age of rail, density of traffic, number of defects found, etc
Approximately two-thirds of the track mileage is tested by Sperry Rail Service under
contractural arrangement with the individual railroads. This company i- ;m outgrowth
of the Sperry Development Company that developed the firsl detector car for the Ameri-
can Railroad Association. I am not sufficiently familiar with the Sperry detector car to
be able to describe its method of operation in detail, It i- mj understanding that the
Sperry car, in general terms, utilizes a combination of the inductance method of testing
combined with ultrasonic units that exph re the rail at the rail ends for head di
Thus the head of the rail is tested for defects for the entire rail length. The remaining
one-third of the track mileage is tested with detector cars that are owned and operated
by individual railroads or contracted for testing service from the AAR Research Center.

Preventing Rail Failures in Track

The AAR road-rail type detector car.

Defect detector pick-up coils at the rear of the road-rail detector car.

Preventing Rail Failures in Track

For detecting defects in the rail web the ultrasonic principle of defect detection has
been found to be effective. There are several different types of units available for doing
this detection. In general these units inspect one joint at a time, the operator of the
equipment slowly moving the crystal over the rail for the length of the joint bar, ex-
amining the web for possible head-and-web separations or bolt-hole cracks. The crystal
generates an ultrasonic wave and transmits it vertically into the rail head and down into
the web. In one type of equipment the reflected sound wave is shown on an oscillo-
scope. The operator is able to see two lines representing the top and bottom of the rail
and when passing over the bolt hole the reflected sound wave is interrupted by the bolt
hole so that a definite pattern is shown for the hole. If there is a crack emanating from
a bolt hole a different pattern will be indicated which will enable the operator to deter-
mine the existence of a bolt-hole crack.

Another type of equipment is quite portable; in fact it can be carried in a sort of
knapsack on the back of the operator. The principle of detection is to pass a crystal
along the length of the rail, transmitting a vertical sound wave into it and the reflected
sound wave is picked up and transmitted to a set of earphones. An experienced opera-
tor is able to detect by the difference in the transmitted buzzing sound in the earphones,
the existence of a crack in a bolt hole or a head-and-web separation. In another type of
unit, the reflected sound wave is indicated visually on a meter.

For several years, the Santa Fe Railway has been engaged in the development of
a road-rail unit to continuously test the rail web for defects by the ultrasonic method.
This unit has a very ingenious coupling method utilizing a column of water to make
the coupling between the crystal and the top surface of the rail, surrounded by a rubber

Road-rail ultrasonic type rail defect detector car developed by the Santa Fe.

Preventing Rail Failures in Track

Crystal coupling unit on the Santa Fe ultrasonic detector car.

suction cup to reclaim most of the water used. Cathode ray viewing scopes show the
reflected pattern continuously for both rails, and two additional retention-type scopes
show an enlarged pattern for each rail joint as it is passed.

It is my understanding that Sperry Rail Service has recently added further equip-
ment to their regular detector car so that it can be used to locate defects in the rail web
by ultrasonic means at the same time that the test is being made for head defects. This
supplemental equipment does not make a continuous record but indicates by a light sig-
nal when a defect is located and the operator of the car can then stop and make a hand
check with a portable ultrasonic unit to verify the existence of a bolt-hole crack or
head-and-web separation.

Several years ago the detector car engineer and metallurgical engineer of the Asso-
ciation of American Railroads made a trip to Germany to see the ultrasonic type of
detector car developed for the German National Railways. This type of car depends en-
tirely upon ultrasonic sound transmitted into the rail with three sets of crystals, one in
the vertical direction, one at 35 deg, and one at 70 deg, to locate defects both in the rail
head and in the web. This car has the advantage of affording high-speed operation as
it can be operated as fast as 40 mph, making a photographic-type record on sensitized
paper of the traces actuated by the reflected sound waves. It was considered that this
type of car had many advantages but that it had two important disadvantages for use
on railroads in the United States. One disadvantage is that the photographic record must
be developed and examined by observers especially trained to locate defects by noting
any deviation from the normal pattern. The existence of a defect would then have to be
communicated to the track men to locate and remove the rail from service. Thus a delay

Preventing Rail Failures in Track

*►

Ultrasonic type detector car developed by the German National Railways.

Three sets of crystal generating and pickup units on the German National

Railways ultrasonic car.

Preventing Rail Failures in Track

Interior view of the German National Railways ultrasonic car.

of several days would occur between the time when the car makes a test and the defec-
tive rail is replaced. Also, there is the possibility that a mistake might be made in locating
the defective rail and, therefore, it would not be removed and replaced with a sound
rail. These two disadvantages and other factors have been considered of sufficient im-
portance so that at the present time only limited use is being made of the German type
of car in the United States. However, I understand Sperry Rail Service does have at
least one car of this type with which it is experimenting and it is possible that further
development may make the detection of both head defects and web defects practical
with this type of unit.

The economic advantages to the railroads that have accrued from the use of defect-
detection equipment cannot be over emphasized. It is difficult to place a dollar-and-cents
evaluation on these advantages. However, an estimate can be made that will serve to
give some idea of the importance to the railways of the use of detector cars for prevent-
ing rail failures from occurring in service. Approximately 200,000 miles of track are
tested each year by detector cars and some 20,000 defective rails are located and replaced
with sound rails before service failure occurs. In one year one large railway had 65 service
failures that were not located by detector cars and from these 65 service failures there
were three train derailments that cost the railway approximately $750,000 in damages
to track, equipment and personal injury. This represents a cost of approximately $10,000
for each service failure. If, therefore, the 20,000 detected defective rails were left in
service until a service failure occurred, it could reasonably be assumed that these service
failures would cost the railroads an average of $10,000 each or a total of $200,0000,000
for the 20,000 defective rails that are detected and removed from track annually. This,

Preventing Rail Failures in Track 55

of course, is aside from the humane considerations of the reduction in personal injury
and possibly death of passengers and employees. The cost of testing 200,000 miles of
track per year is on the order of S4.000,000 and this cost would perhaps be increased
to approximately £6,000,000 if the cost of operating the ultrasonic equipment for testing
for bolt hole cracks and head and web failures is added thereto. Thus it is apparent that
the railways have benefited greatly not only in safety of train operation but also in
economy of train operation by virtue of preventing rail failures in track through the
use of defect-detection equipment.

Welded Railroad Bridges*

By J. E. SOUTH

System Engineer — Structures, Pennsylvania Railroad

It is interesting to note that the first all-welded bridges in this country and possibh
in the world were railroad bridges constructed in 1928.1 The first bridge was a single-
track through truss with a span of 135 ft built to carry the railroad over a canal at
Chicopee Falls, Mass., for the Westinghouse Company. The second bridge, also built in
1928 at the East Pittsburgh Plant of Westinghouse, is an open-floor single-track through
girder span of 55 ft. The floor system of this bridge is made up of rolled beams. No
connection angles are used — the ends of the stringers are fillet welded to the floorbeams
with a tension plate carried through the floorbeam web, making the stringers continuous.
The floorbeams are fillet welded to the girder webs. The girders are made up of Y% by
54-in web with three cover plates for each flange. One cover plate of each flange runs
full length, and the other plates are cut off as would be done with a riveted bridge.
The shortest cover plates are Yz in by 1 ft 4 in, the next plates are J^ in by 1 ft 2>l/$ in,
and the full length plates are % in by 1 ft 2l/2 in. Lateral bracing is welded directly
to the bottom flanges of the girders and floorbeams.

Another early all-welded bridge is a two-track bascule span built in 1935 for the
Florida East Coast Railroad. Plans and specifications for this bridge were prepared by
the railroad and the bridge was fabricated by McClintic-Marshall. J. M. Wolfe, chief
engineer of the Florida East Coast, advises that the bridge was designed for Copper E 55
loading, and in the 26 years the bridge has been in service it has carried each day an
average of 10 passenger trains with 16 cars each and six freight trains with 60 cars each,
or a total of 150,000 trains and about 10,000,000 freight cars. He further states that they
have experienced no difficulty with the \ve!ded work and have had no weld failures.
In their experience they have had more corrosive action from brine drippings from
refrigerator cars on rivet heads than en welds similarly exposed. In fact, he reports that
there has been practically no corrosion on weld metal.

In addition to these early bridges, welding has been employed extensively from the
early 'thirties by many railroads for the strengthening and repairing of steel bridges.

Welded railroad bridges in a number of ways are preferable to riveted constructirn.
the mest important being the matter of initial cost. To be certain of this point, a few-
years ago the Pennsylvania Railroad took bids on a deck-plate-girder bridge with a 65-ft
span. One span was designed for riveted construction and one span i< r welded construc-
tion. The riveted span weighed 25 percent more than the welded. The low bid was 16
cents per pound for the welded bridge and 17 cents per pound for the riveted bri Ige
Since that time similar welded bridges have been purchased fi t as low as 13.4 cents per
pound. The economy of the all-welded bridge has definitely been established.

The clean surfaces of the welded bridge make painting easier, and the paint last
longer. Rivet heads and edges of plates and angles are usually where the paint film first
breaks down as it is the thinnest at these points. Dirt that is often trapped by tivel
heads will many times be washed off the flanges of the welded bridges by a heavy rain
The lighter weight of the welded bridge often facilitates erection.

•Presented before the Railroad Sessions of the Transportation Engineering. Conference of the
American Society of Civil Engineers in Detroit, Mich., October 8-11, 1962. This papa «.i-

ten features in the program cl the Railroad Sessions, on October 9, developed by the Boat
Direction of the AREA.

1 Arc Welded Steel Frame Structures — Gilbert D. Fish McGraw-Hill.

Welded Railroad Bridges

""M*'n Stt ™
All-welded bridge built in 1928 at East Pittsburgh, Pa.

. ■ ■■■■

Removing old cover plates in connection with cover-plate replacement work,
Pennsylvania Railroad's West Philadelphia elevated line.

Welded Railroad Bridges 59

The proven economy of all-welded bridges and the many other advantages will
force their eventual adoption by most railroad bridge engineers. However, before that
happens, the engineers must be satisfied beyond any doubt that welded bridges will be-
as reliable from a safety standpoint as the riveted bridges have proven to be since the
turn of the century. The development of reliable riveted bridges was accomplished only
over a considerable period of time. In the 1870's and 1880's railroad bridges were failing
at the rate of about 25 per year.2 It was only the employment by the railroads of engi-
neers specializing in bridge engineering who through research, scientific design, careful
detailing, thorough testing of materials and inspection developed the reliable riveted
railroad bridges we have today.

It has only been since the end of the second world war that study and research in
connection with the brittle failure of a number of the more than 500 all-welded Liberty
Ships in service by January 1943s that the different requirements for welded and riveted
structures was understood.

When brittle failure occurs the crack usually propagates very rapidly across the
entire width of a plate. When plates are welded together they act as a single plate,
whereas in riveted work plates fastened together by rivets act more as individual plates,
thus limiting the extent of the fracture. In addition, thicker plates are generally used in
welded structures. Thicker plates are more sensitive to brittle fracture, from both
geometrical and metallurgical factors.4

In addition, the notch effect produced by the restraint of welded connections, the
notches due to imperfections in the welds, the residual stresses caused by weld shrinkage,
and many other factors, contribute to making welded structures more prone to brittle
failure than riveted structures.

To prevent the rapid propagation of a brittle fracture the steel must yield plastically
in the presence of a notch. The tendency toward failure in a brittle manner increases as
the temperature decreases and the rate of loading increases. Studies of welded ship fail-
ures by the Ship Structures Committee6 concluded that a steel that was not notch sensi-
tive at low operating temperature was required, resulting in the standard specification
for structural steel for ships, ASTM A 131.

It is interesting to compare the requirements for ship steel with the requirements
for welded bridge steel covered by ASTM Specifications. A 36, A 373 and A 441. Cer-
tainly the speed of application of load and low range of operating temperature is as
severe for a railroad bridge as for a ship. One main difference in the structures is that
it is more difficult to eliminate geometrical notches in ship construction, and the large
size of the ship increases the sensitivity to brittle failure. Since tension flange plates of
welded bridges are usually 2 in or more thick ("which inddently i> the greatest thickness
covered by the ship steel specification) the following comparison is confined to thick-
nesses in this range:

CnEMicAL Requirements"

A 131
Grade C° A 373 A 441

C max % 0.23 0.26 0.26 0.22

Mn % 0.60 to 0.90 0.50 to 0.90 0.80 to 1.20 1 2S max

Si % 0.15 to 0.30 0.15 to 0.30 0.15 to 0.30 0.30 max

V min 0 0 0 0.02

* All requirements shown tased <>n ladle analysis.

b Plate steel produced to requirements of Grade C '-hall he made with fine grain practice . . .

3 Railway .Age Centennial 1

3 Control of Steel Construction to Avoid Brittle Failure Welding Research Council, page 31.

♦Control of Steel Construction to Avoid Brittle Failure Welding Research Council, pagi

■Welding Research Council Bulletin Xo. 16.

Welded Railroad Bridges

Another view of cover-plate replacement work, West Philadelphia

elevated line.

Welding new 18- by iy2-in. plate.

Welded Railroad Bridges 61

Tensile Requirements

A 131 A 373 A 36 A 441

Tensile strength 58,OCO to 71,000 58,000 to 75,000 58,000 to 80,000 63,000 min

Yield point min 32,000 32.000 36,000 42 0:Omin

Elongation in 8" min ... 21% 21% 20% 19%

Elongation in 2" min ... 24% 24% 23% 24%

There is no reason to believe that A 36 steel will have more notch toughness than
A 373, and it is known that A 373 steel has failed by brittle fracture while field welding
a deck plate to the top flange plate of a welded railrcad bridge girder. There did exist,
however, a severe notch because of faulty construction procedure in cold weather. There
have also been brittle failures in service of A 441 steel under low temperature conditions.

It has been shown that the transition temperature of thicker plates is higher than
thinner plates of the same heat. Ship steel has been limited to 2 in. in thickness, and all
steel over 1 in thick must be Class C made to fine grain practice. Thicknesses are
permitted up to 4 in for steel for welded bridges, and in the case of A 373 and A 36,
an increase of carbon to 0.27 percent is permitted. In the case of A 441 steel there is no
minimum silicon specified, implying that killed or simi-killed steel is not required. There
is no minimum manganese requirement, so presumably if the manufacturer can find
some other strengthening material to enable meeting the tensile requirements, it could
be used regardless of the effect on notch toughness. It is not the average steel that is
produced under the specification that is of most concern to the bridge engineer, however.
It is the possibility of obtaining the worst possible steel that would still meet the
specifications that is a disturbing factor.

There has been no standard set for notch toughness requirements for steel for
welded bridges, but there seems little reason to believe it should be much less than
that required for ship construction. The rate of load application on a railrcad bridge
is probably more severe than for a ship and the thicknesses used are much greater. It
would seem logical that steel for important tension elements for bridges in thicknesses
over about 1 in should be made with fine grain practice as is the case for ship steel
over 1 in.

To the present time there have been few fatigue failures in railroad bridges in thi>
country except for connection angles, secondary members and members carrying high
live load and low dead load and subjected to high secondary stresses.

Since it has been demonstrated that certain details, such as short length cova
plates whether riveted or welded and other stress-raising details, can reduce the fatigue
strength at 2 million cycles of zero-to-tension stress range to as low as 10.000 psi," it i-
quite possible that the high impact allowances that have been used in the past also pro
vided a safeguard against fatigue failure.

With more accurate measurements of impact hading to a reduction of the Impart
loads combined with the higher stresses to be permitted with the A 16 Steel) probably
more consideration should be given to the effect i I fatigue on the shorter spans (under
60 ft) and where the full effert is obtained by loading s single track, In any case, in
the zero-to-tension stress range at 2 million cycles, members fabricated of bigh-strengtb

9 Prof. W. M. Wilson, Univ. "f Dlinois Akl \ Pi linj Mlnob

Bulletin Series No. 382.

Welded Railroad B ridges

PONTOON

ABUTMENT

LONGITUDINAL SECTION

Details of welded float transfer bridge, Norfolk Portsmouth Belt Line
Railroad, Norfolk, Va. See also drawings on opposite page.

i&'» Flame Plate
thickness \

inq
icKi

Edgeofplate-^

/—Edge of plate

GIRDER FLANGE JOINT

Welded Railroad Brid ges

a-a DETALS-OUTSHORE END

SLOPE 2nni-» * 60* *

WEB PLATE JOINT

MID -SECTION

-10".

60' ^;* %

NOSE SECTION

GIRDER SECTION

64 Welded Railroad Bridges

steel do not show any significant increase in fatigue strength over A 7 or A 373 steel,
so for certain short-span members no advantage can be gained by use of higher
strength steels.

On spans of about 60 ft and longer during the passage of an entire train the load-
ing will increase from zero to maximum and then fluctuate from about half maximum
to maximum. In this case fatigue becomes of considerably less importance. In the case
of longer spans (over 100 ft) and members receiving the maximum load from two
tracks, it is quite possible that the number of cycles of maximum load would be so low
that fatigue effect would be unimportant.

A study of Variable Stress Cycle Fatigue is now being carried out at the Ecole

Polytechnique, Montreal, Canada, under the guidance of the Fatigue of Welded Joints

Committee of the Welding Research Council that should be helpful in evaluating the
fatigue requirements for railroad bridges.

The design formulas in Table I of the AWS Specification for Welded Highway and
Railway Bridges are based on 2,000,000, 600,000, and 100 cycles of loading. There seems
to be need for a study that would develop the number of loading cycles that should
be considered for various spans and leading conditions.

It might be that the simplest method of providing for fatigue would be to assign
lower allowable unit stresses for certain span lengths and loading conditions. Certainly
the actual ultimate carrying capacity of a dynamically loaded bridge cannot be deter-
mined by having equal stress in all members as would be the case in a statically loaded
structure.

Fatigue cracks are of more importance in a welded structure only because of the
notch effect of the crack leading to possible brittle failure.

Regardless of some of the above uncertainties many all-welded railroad and high-
way bridges have been built and are functioning satisfactorily. In the New York Times
of August 21 there was an article about the failure of a new $10,000,000 welded girder
bridge in Melbourne, Australia, emphasizing the need of careful design, good workman-
ship and use of the proper steel for the job.

Design should follow the requirements of the American Welding Society Specifica-
tions for Highway and Railway Bridges.

In addition, the engineer should minimize the possibility of having a geometrical
or metallurgical notch effect in tension members. For instance:

a. Do not weld to tension flanges of beams or girders.

b. Avoid unnecessary splices in tension flanges. Have all such splices and others
in regions of high stress radiographed to insure weld soundness.

c. Grind flange splices smooth and to a rounded contour.

d. Inspect flange-to-web welds by dye penetrant or magnetic particles or other
satisfactory method to insure against weld cracks.

e. Use full penetration welds in important web-to-flange connections.

f. Carefully inspect all welds to see they are smooth and without unfilled weld
craters.

g. Edges of flange plates should be smooth and without nicks caused by handling
during fabrication or erection.

Welded R ailroad Bridges

Washout of timber trestle, Crawfordsville Branch, Pennsylvania Railroad.

Typical welded steel replacement trestle.

null. 573

Welded Railroad Bridges

Typical details of welded trestle.

h. Be certain that procedure for field welding will not produce undue shrinkage
stresses or result in stress-raising notches.

i. Arc strikes should be removed and voids filled with weld metal and surface
ground smooth.

Use the thinnest plates possible consistent with design requirements. For instance,
if design considerations permit, it would be preferable to use a 24- by 2-in flange plate
rather than a 16- by 3-in plate, giving the same area. Thinner plates have better notch
toughness than thicker plates from the same heat.

Steel for important tension elements can be ordered made to fine grain practice.
It has been estimated that this would add 1.3 to 1.75 percent to the cost of welded
bridges.7

If extra-thick tension plates are needed it would be desirable to specify that they
be normalized to lower the transition temperature. Normalizing will have greater effect
on steel made to fine grain practice.

Railroad bridges in regions of extremely low temperatures should be given special
consideration in regard to transition temperature of steel in tension elements.

If possible, select a type of structure where the failure of one member will not
result in the instability of the entire bridge. For instance use a series of longitudinal

7 Control of Steel Constructon to Avoid Brittle Failure — Welding Research Council — page 130.

Welded Railroad Bridges 67

stringers or girders instead of just one pair carrying one track independently of adjacent
tracks.

There are other things the engineer could do, such as limiting the allowable stress
in short spans which are likely to receive a large number of stress cycles of relatively
high intensity; avoiding high-strength material in these circumstances as such steels
show no increase in fatigue strength in fabricated form, and being certain that the
welded design has not introduced high secondary stresses.

There is no doubt that by scientific design, careful detailing, use of proper steel
for the job, and thorough inspection that all-welded railroad bridges will be as reliable
as the riveted bridges and will be built at considerable savings to the railroads
and others.

Advance Report of Committee 30 — Impact and Bridge Stresses
Report on Assignment 6

Concrete Structures

Collaborating with Committee 8

P. L. Montgomery (chairman, subcommittee), J. YV. Davidson, W. E. Dowling, C. E.
Ekberg, Jr., N. E. Ekrem, J. A. Erskine, J. F. Hoss, Jr., R. E. Kuban, K. H.
Lenzen, C. V. Lund, James Michalos, N. M. Newmark, L. P. Nicholson, M. Noys-
zewski A. L. Piepmeier, E. D. Ripple, A. P. Smith, C. A. Still, F. W. Thompson.
J. R. Williams, J. D. Woodward.

Your committee presents as information abstracts of two reports on investigations
undertaken for the committee by the research staff of the Association of American
Railroads.

FIELD INVESTIGATION OF FLORIDA EAST COAST PRESTRESSED
CONCRETE BEAMS*

This report covers the testing and analysis of two 30-ft 6-in long spans in ■
ballasted-deck precast, prctensioned concrete double-track bridge over Cypress Creek
Pompano Beach, Fla. East test span is comprised of sh 2 it 3 in wide l>y 2 u 6 in
deep rectangular beams per track and are identical except thai the beams In one ipan
had longitudinal recesses cast along their sides near the top which formed kiy\v;r

•An abstract of Report No. ER-21 issued by the Research Department VAF thr full

report can be obtained from the director of enginccrine research, A\K, 1140 9 I IfO It

Prestressed Concrete Structures

tilling with cement mortar. The purpose of the invesligation was to compare the static
and dynamic effect on spans with and without shear keys before and after transverse
post tensioning under the passage of a test train operating at speeds up to nearly 70
mph. The test train was composed of a two-unit diesel locomotive, five 50-ton and two
70-ton loaded hopper cars and a caboose.

Either shear keys or transverse post tensioning is effective in distributing the load
across the deck. An improvement in load distribution was noted for the span with shear
keys and after post tensioning. Before post tensioning, in several instances, the recorded
static strains in various beams exceeded the calculated values. For the span without
shear keys, the outside beam carried negligible load until after post tensioning. On the
other hand, the outside curb beam and the other five beams in the span with shear
keys carried approximately the same load before and after post tensioning. The post
tensioning caused some of the load to be transferred to the span in the adjacent track.

Comparing the spans, the one with shear keys had the lowest range of impact
values after post tensioning. While the maximum recorded total impact for the locomo-
tive did not exceed the current AREA specifications, it is interesting to note that the
increase in strains with increase of speed was appreciable only at speeds above SO mph.

FIELD INVESTIGATION OF SOUTHERN PACIFIC COMPANY, TEXAS &
LOUISIANA LINES, PRESTRESSED CONCRETE GIRDER SPANS*

A test was conducted in 1961 to determine maximum strains, static effects and load
distribution in two spans of a ballasted-deck single-track bridge near Houston, Tex.
Both spans have cast-in-place concrete decks supported by precast concrete girders

* An abstract of Repurt No. ER-25 issued by the Research Department, AAR. Copies of the full
report can be obtained from the director of engineering research, AAR, 3140 S. Federal St., Chicago 16.

Prestressed Concrete Structures 71

having the same I-shaped cross section. One span contains four 30-ft long pretensioned
girders while the other has five 55-ft long girders which were first pretensioned and
later post tensioned. The test loading consisted of two- and three-unit diesel li comotives
operating at speeds up to 20 mph.

Two different approaches were followed in the determination of calculated strains.
In one case, the basis was to assume that the entire cross section acted as a composite-
unit. In the other case, one girder and the corresponding portion of slab was assumed
to act together. The maximum recorded strains occurred at the maximum speed but
they did not exceed the calculated values regardless of which approach was followed.
The study of vertical strain distribution was dependent on the same assumptions
used in calculating the neutral axis. On the assumption that the entire cross section
acted as a unit, the neutral axis based on average recorded values was below that
calculated for the shorter span and approximately the same for the longer span. Com-
posite action existed between the precast girders and the cast-in-place decks of both
spans. In regard to transverse distr.bution, the two inside girders of the shorter span
carried approximately 63 percent of the total load. Each of the five girders in the longer
span carried nearly equal loads, and the interior girders also carried approximately
60 percent of the total load.

Advance Report of Committee 16 — Economics of Railway
Location and Operation

Report on Assignment 4

Potential Applications of Electronic Computers to Railway

Engineering and Maintenance Problems in Research,

Design, Inventory, Etc.

Collaborating with Committees 11 and 30, and Informally with the Railway
Systems and Procedures Association

F. Wascoe (chairman, subcommittee), L. P. Diamond, W. J. Dixon, G. B. Dutton, Jr.
S. B. Gill, C. A. James, T. J. Lamphier, R. J. Lane, A. S. Lang, M. B. Miller.
V. J. Roggeveen, G. Rugge, G. S. Sowers, J. J. Stark, Jr., C. L. Towle, T. D.
Wofford, Jr.

Your committee presents as information two reports on the use of computers in
operations research on the Canadian National Railways. One of these reports is entitled
"Train Performance Calculator" and the other, "A Computer Simulation of Railroad
CTC Operations."

Train Performance Calculator

By CHARLES SANKEY

Operational Research Branch, Research and Development Department,
Canadian National Railways

The Canadian National Railways' Train Performance Calculator (TPC) is a method
of using an electronic computer to simulate a train's movement along a railway lint'
The results include a graph showing the train's speed all along the line, and tables of
running times, station-to-station times, locomotive power factor and other useful
information.

The TPC program was developed by the Operational Research Branch in early
1960 at the request of the Transportation Department. Since then it has been exten-
sively used, particularly by the Operating Department. Among the questions investigated
by the use of TPC are:

1. To find the time any particular train will take to cover a .subdivision (or
several subdivisions). This is particularly useful in train scheduinp.

2. To find the amount of power required (i.e. how man) diesel units) to make
a run in a certain time.

3. To determine how much time is spent, and how much extra fuel is used, ii t
train makes an extra stop.

4. To find out what effect a speed restriction, or the removal of ;■ speed restrii
tion, will have on a train.

5. To estimate the effects of rolling-stock or locomotive improvements
rcl'.er bearings instead of solid journals) on schedules and fuel D DSUmptii n

TPC can be used just as easily for a track which has not been built, or i<t > nun
which does not exist, as it can for an actual train on an actual track. This is of the
greatest use in planning new lines, for example, where it is necessary to decide what th<

74 Train Performance Calculator

best route or the maximum grades and curvature should be. Approximate rules and
guesses can be replaced by detailed and accurate information.

In the past few years, several methods have been devised by different railroads
to perform train performance calculations on an electronic computer. In the CNR pro-
gram, particular emphasis has been placed on the ease with which it can be used and
in simulating actual conditions as accurately as possible. Among the special features are:

1. The length of the train is fully taken into account. This is important firstly
because the effective grade under a long train is different from that under a
short train, tending to smooth out variations in grade, and secondly because
speed restrictions generally apply to the whole length of a train — for example,
restrictions through turnouts. The other case, where a slow order applies only
until the engine has passed a point such as a grade crossing, is also correctly
treated.

2. The computer uses an automatic "look ahead" method to determine when the
train should apply brakes in order to comply with a lower speed limit or
come to a stop.

3. The effect of track curvature in retarding a train is allowed for in the com-
puter by using the relation that each degree of curvature is equivalent to a
certain increase in gradient.

4. Locomotive tractive forces and train resistance are given to the computer in
the form of tables as a function of train speed. This is much more flexible
than assuming a constant horsepower or using Davis-type formulas for train
resistance.

5. The actual calculations are done by a step-by-step solution of the differential
equation governing the train's motion, expressed in terms of energy rather than
acceleration. The effect of this is that a larger interval between successive steps
can be used. We have found that a lS-sec interval using the energy equation
gives as much accuracy as a 5-sec interval using accelerations. The result of
this is to reduce the number of calculations required for a run.

6. The "velocity profile" (i.e., graph of speed versus distance) which is produced
as an output shows very simply and clearly such things as the effect of slow
orders, the minimum speed up the ruling grade, and the relationship between
actual train speed and the speed limit.

7. The quantities entered on a control card and therefore completely under the
operator's control, include:

(a) The effective value of "free-fall" gravitational acceleration, modified by
rotational inertia, which varies from train to train.

(b) Braking deceleration.

(d) Length of train.

(e) Weight of train.

(f) Time interval for calculating successive steps.

(g) Distance interval for detail output and for the velocity profile.

8. Preparation of the input to the computer is very simple. One set of punched
cards contains tractive forces and train resistance, another gives speed limits,
a third has elevations and curvature at arbitrary points along the track (usually
where the gradient changes), and a last set gives temporary speed orders (slow

Train Perform ance Calculator 75

or fast), stations for which summary outputs are required, and train stops.
An interesting feature of the last three sets is that they are reversible. In order
to run a train in the opposite direction, the cards are merely turned around
and read into the computer backwards, without being sorted or rearranged.
This is accomplished by plug-beard wiring in the computer.

Although there are always differences from the ideal in actual train operation, and
no two train runs will ever be identical, the TPC results have shown remarkably good
agreement with measured train times.

The Transportation Department has used the program to prepare charts for all
main-line subdivisions, eastbound and westbound, giving running times as a function
of weight/power ratio for passenger trains and for freight trains with various values
of average car weight. These are being distributed to the Regions, and it is believed
that as they become more familiar to management and operating personnel, more uses
will be found for them and for the greatly improved information which they contain.

APPENDIX A— EQUIPMENT
IBM 650 with:

Basic alphabetic device

2000-word drum storage

On-line IBM 407 tabulating machine

The program is at present being rewritten to make use of the IBM-7070 computer
which has recently been acquired by the CNR. Various additions and modifications will
be incorporated at the same time.

APPENDIX B— MATHEMATICAL METHOD
Units

For simp'icity and precision in operations within the computer, the following
quantities have been used:

Distance .V miles

Time T minutes

Speed V miles per minute

Acceleration A m'les per minute1

Energy Potential P (miles per minute)

For machine use ( but not tor input and output), all variables have been expressed
in these units. For example:

I I- in lb ui

\r = acceleration due to tractive effort = — ■ . . • ,, G

tram weight m ll>

. . , Resistant e in lb w i

An = deceleration due to resistive forces = ; — : . . , . — r. ■ (,

train weight in lb

Pit = Energy potential due to height (including curve effeel
= average height of train in miles X G

In these examples, G is the gravitational acceleration constant, modified to lake
account of rotational inertia.

76 Train Performance Calculator

Full-Power Operation — i.e., Below Speed Limit

A step-by-step integration procedure is used, with an approximately constant time
increment.

Let DT, DX, DV, DPn denote increments of time, distance, speed and height
potential, and let DCT be the basic time interval. Suppose that after any step the train's
position and speed are X and V.
Then:

DX = (V + l/2 DoV) DeT, where D„V is a first approximation to DV (see below).
Calculate DPn corresponding to DX, from the elevation card input.
Calculate Am = Ap — AR for a speed V-\-l/2 D0V, by interploating in the table
of tractive force and resistance.

Am DX — DPb

DV — v+y2D0v

_ DX

DT~ v+y2DV

Provided that no braking lines or speed limits are violated, these increments are
then added to the old variables and the process is repeated, using the value of DV as
the new value for DaV.

For the first step after release from a speed limit, or from a stop, there is no
available value for D0V, and so it is calculated:

DoV =(AM — AG) DcT, where Ao is the acceleration due to the gradient under
the train.

Train Performance Calculator

••

T3TD '

C « O

re o "V

<£

«/>

,«t-5 c

4*.

u ©

•i

O U A

<ANt»«<MnN«prtHN4<#a4rth-o»H«oo»

*ss

f»-#**d«0»*^-*£W«>l^.#oj?>f*o>^©«**\©0>«n

V9 •

z>v>

C t/)

— OQC

-" E c

vctl

<UJ

Kzn.

t» z>

oouxx

aC c

•OCT

W « re

•

atri

•- Ui

XUJ

•2E -

ujza.
uxr

« P »)

•i •

C o

I-

o C ll

« i<

4f \t

\m d tl

*(<

1:1

o w

+J to

O i •

a a

w — u

° o °

a. 2 j=

4> -^ n_^

O «o
• ZtU

« rt « * «* ~> m o- < ir o

u-cu

<M ©> o>> «o • •* «A *\< *> «■«

■-J «

oxa

• • • • •• •••••

Sog;

_j •/>

r\ «n «-> o m o* «n o «> e> •>

hhn r> *i «f\ r- «o> o **

j.-

•
ox

r-4 r4

rent
d o o

1*1

xa •

*\ r- o o

• • • •

• • • • • • •

72 ■£ <->

«az

r- • r» • to ■€ * «o « m <C

E « °

aOT >,

—•

i*i*

'C n-°

xv>

H w —

•z
a •

•• 5 2

U V .,

«^> re

<u>

-IK

UiX

_ t» W
.« WO

T3 ac

w X lu a

re <" t>

Ui tt

u a oc_#o z o < a: >- £

J=T3 X

UJ BE U

v2 v

ae<

u uj ir >- z

jEo

BMOMXU jaillO J">K

H.c bo

»-

r «/i >w)OZ<<

OUlLhlOfCOUX

<c 3v

jiflMLwSNkjhU*!*! WJ

<z>ij<ii.>f3»j»)Kq

&. z 3

jj ~ *>

«o

HhJIIIMi^U<

M E J

3 r C

•36

<OQkJ<M<UJ

5§3

3*5 -

o « i2

X-i

•

eoooooooooooooooooopoo«v«««

>» bflO

c E iS

5 3 tn
Ijg

•*©«©-#r>#-«M«o.#rMr^rx«D.©«>.->.*«>c><.*^.tfr--.#*.

■J

NNHOOfre

■S«)f-f-^>-OrfN*>-*^«^«^ — — •

»-_)

w ^

bo tn"°

•

• ••«•••••••••••••••••••••

.^u c

o«\ • m» «» s " . ^ »- «m r- — tr^^<r^*>»^r~^»'<ri«M«^<»

^ 3 O.-

.3 o re
ere-"

»-«»<*<iMr»'**'#tr*\«<«r>-f-«»»»'©©«-,'«<»«w

«•

Train Performance Calculator

SU3D

VISON

IL UP KAII'

KLKI-yr<

M«N(.t UU

TRAIN NO. /c«/7 Ar**r+>/tfJb

TRIP M/P
MILEA6C LOC

stat; on

ELAP.TIME
HRS.MINS.

SPEED
M.P.H.

COMP.GR
%

POWER U. EXP. 8RA<INo

, ENERGY ET.ERGY ST, TO ST

P. AVAIL.* HP HR.S HP hRS. M1NS.

5 0*00

129.58

EDSON

S 0.00

12^.58

EDSON

0.25

129.33

01.0

25.6

.12-

0.50

129.08

01.5

33.4

.38-

0.75

128.83

01.9

35.0

.36-

1.00

128.58

02.4

35,0

.34-

1.25

128.33

02.8

35.0

.27-

1.50

128.08

03.2

39.9

.20-

1.75

127.83

03.5

43.4

.21-

2.00

127.58

03.9

46.5

.32-

2.25

127.33

04.2

49.4

.35-

2.50

127.08

04.5

52.0

.35-

2.75

126.83

04.8

54.4

.35-

3.00

126.58

05.0

56.5

.37-

3.25

126.33

05.3

58.4

.38-

3.50

126.08

05.5

60.0

.38-

3.75

125.83

05.8

60.0

.38-

A. 00

125.58

06.0

60.0

.39-

4.25

125.33

06.3

60.0

.39-

4.50

125.08

06.5

60.0

.40-

4.75

124.83

06.8

60.0

.40-

5.00

124.58

07.0

60.0

.40-

5.25

124.33

07.3

60.0

.19-

5.38

124.20

YATES

07.4

60.0

.08-

5.38

124.20

YATES

07.4

60 322 9 7.6

5.75

123.83

07.8

60.0

.43-

6.00

123.58

08.0

60.0

.38-

6.25

123.33

08.3

55.9

.34-

6.50

123.08

08.6

49.0

.06

6.75

122.83

08.9

41.0

.16

7.00

122.58

09.3

40.0

.05

7.25

122.33

09.7

40.0

7.50

122.08

10.0

40.0

.05-

7.75

121.83

10.4

40.0

.46-

8.00

121.58

10.8

44.4

.53-

8.25

121.33

U. 1

48.1

.29-

8.50

121.08

11.4

50.7

.22-

8.78

120.80

WOLF CREEK

11.7

53.0

.30-

8.78

120.80

WOUF CREEK

11.7

58 459 170 4.2

9.00

120.58

12.0

54.9

.35-

9.25

120.33

12.2

56.9

.36-

Fig. 2 — Detail output. Since the summary cards have not been sorted
out, they also appear in the print-out. The headings referring to the details
cards are: Trip mileage, Milepost location, Elapsed time, Speed, Gradient,
compensated for curve effect.

Train Performance Calculator

VELOCITY PROFILE
SUBDIVISION XJ

OATF t/9/i/

4_
'3

„-

0
I o

D
0

c ■>

J
I

Z 0

X
Zo

Ul
Itl
or

111

c<s>/»-cru-f-rrurr(-arrrur<rcr'rur^furtfurcrurrrur<rur-C'«rf<--»rcrKr''--'

Fig. 3 — Velocity profile. The figures "1" refer to the train speed. "2"
to the speed limit. Where the train is travelling at the speed limit, a "3"
is printed.

Train Performance Calculator

€,°iod8l
TOT

•)*9drr>S
V6

L'6 „

BUB30£

1*6

TBU^ag

TOOOMO 01 c

JTnpexo

B3BU0TJ

qsn^BJinM
T .

a -

a ^

J3_*_

a a

a 3 a a

a a

— ; — cr-

? a

o o

^ w

.5 c

re1"*

Sis

go
u

»%
is

"■^ c
c ^
cd u G

w O «*>
• _. w

gE,rt

U*2

S h'*f,

tn.E
re C V

3 re

E « <u

bO-O.S

.Sew

C w

Egv,

|.2E

.5 o

ET3 W.
OwJg

t: « «

J3 * „
Owe

■ >.s

o c

Train Performance Calculator

C 1- !/)»*-

3 re >, o

- £ C

X to re re

--c

M re h

E w n

> 2 «

•GX w'O

&6*"'S

^•s§

•-* re _c »h
u o 5 u c

•2 re p c«c
0- 53 m « c

o c

re '

c ac ™ a

re ;,, -"" en

4_j

x
Eh

re
x

in
II

>
re

B
g

V£

X vo

<N1

(Nl

•■

+-j

M
II

L)
II

u
U

>
a)

■y;

■_

re
-
re
A

7.
go

re
>

0
A

'53

_re
•-
O

a
0

4J O

bfl

3
E
'3

re
X

0
~

WV
I3

■

■■

.5*x

„_;

Et,

>
0

A Computer Simulation of Railroad CTC Operations*

By C. J. HUDSON

Operational Research Branch, Research and Development Department

Canadian National Railways

A. INTRODUCTION
Background

The Canadian National Railways are currently implementing a program of Cen-
tralized Traffic Control on 40 main-line subdivisions, a total of 4500 miles of single track.
Since the cost of installing CTC is high ($50,000 per siding or approximately $1 million
per subdivision), it is essential to determine the most economic number and locations
of signalled sidings which will adequately handle the current and anticipated traffic.

In the past, upon a proposal to install CTC, the number of sidings required has
been based on the grid system, their location determined largely by local operating per-
sonnel and the suitability of the configuration evaluated by a manual redispatch. This
process of drawing by hand the trains' time-distance graphs and planning the meets as
would an experienced CTC dispatcher is time-consuming, and any quantitative measure
by which one siding configuration may be compared to another entails much further
work. The time taken to redispatch one day's train over a subdivision runs into several
man-hours.

The Operational Research Branch was asked to develop a technique which would
enable the Transportation Department to evaluate various proposals for siding con-
figurations easily and quickly, and to compare these alternatives quantitatively.

Many persons have been associated with the development of this project from its
beginning — in particular, J. M. Davey, KCS (Quebec) Ltd.; I. H. Cole, now with A. D.
Little Inc., and members of the Transportation and Maintenance Department.

Program Design

Various approaches to the problem were made, including linear programming and
combinatorial analysis. In an analytic formulation the complexity of the problem is
greatly increased by introducing the siding spacing, siding capacity, and train speed,
length, density and departure times as true variables. Yet it is essential that each of
these remains variable if the representation is to be realistic and the solution of prac-
tical value. From these considerations it was decided that the most suitable approach
would be a direct simulation using a digital computer.

By direct simulation is meant that an attempt is made to create in the computer
a representation of the situations that occur on a subdivision. In the computer represen-
tation of the CTC subdivision and the train traffic, the program moves the trains,
predicts meets, decides the trains to be delayed, arranges meets at the sidings and
imposes on the delayed trains the delays they incur in practice.

The use of a computer to represent a subdivision and its traffic is closely analogous
to that of a CTC dispatcher's console, which is itself a direct simulation of the track.
Here the dispatcher observes an indication of the presence of a train in a certain block
of track by a small light in the appropriate location on the console. The equivalent in
the computer might be the presence of a number, identifying the train, in a specific
memory location.

Presented at the 1962 Joint ASME-AIEE-EIC Railway Conference, Toronto, Ont., April 10, 1962.

A Computer Simulation of Railroad CTC Operations 83

There were two stages to the development of the program. First the objectives
of the program had to be translated into a logical set of operating rules. Computer
techniques were then developed to represent and move trains. The objective of the pro-
gram is to "duplicate" good dispatching of trains over a subdivision. A dispatcher's
objectives can be stated simply:

1. Scheduled traffic should maintain its schedule as far as possible

2. In delaying trains, priorities must be recognized.

3. The overall delay to all trains should be as small as possible.

These objectives imply that:

4. Future meets must be anticipated.

5. A decision must depend on concurrent consideration of (1), (2) and (3)
above.

A set of decision-making rules now used in the program was the result ol many
discussions with dispatchers and operating personnel, analysis of CTC train charts, and
trial-and-error methods using the computer. One of the difficulties was to formulate a
set of rules which apply to any subdivision. Dispatchers, with considerable experience
of local traffic, have intuitively created for themselves a set of rules which are suitable
for the efficient movement of trains on their particular subdivisions; and with highly
repetitive patterns of movement, experience is one of the most useful factors of decision
making. Such a rule might be "If No. 1 doesn't overtake No. IS by siding X it will
arrive late at the terminal."

There are many ways of building a program to move trains, detect meets in ad-
vance, implement a decision once it has been made, and to initiate and terminate trains.
These are matters of technique and, in this case, the prime considerations in determining
the type of program were speed and convenience of operation, using the features and
capacity of the available 4000-word IBM 650.

B. DESCRIPTION OF PROGRAM
Data Preparation

When a decision is taken to study a subdivision prior to CTC installation, the
Operating Department first chooses one or more samples of actual traffic which are
considered suitable for evaluation of the siding configuration, the traffic being either
typical or as heavy as will be encountered on the subdivision in the foreseeable future.
These samples usually consist of a week's trains, though any lesser time period ma\
be used. If the installation of CTC indicates any changes in schedules, departure time]
or length of trains, these are allowed for in preparing the data The data which are
punched on cards completely specifies each train and includes such items as the running
characteristics, schedules, number of cars and tasks enroute.

Track and train running-time data complete the input to the simulation. Tin
information is that needed completely to specif] the CTC configuration, namely, the
location of signalled sidings, electric lock sidings and junction-, and the cai capadt]
and switch arrangement of sidings.

Instead of calculating the siding-to-siding running tfnv cfa nam pp

these times are interpolated from a series of time-distance tables, which are pari ol the
results of another simulation, the Train Performance Calculai 01 or TP( I I train
of given characteristics, this program simulate- its movement QVtt ol track,

taking into account elevation, curvature, speed limit- and temporary -low orders Cal

S4 A Computer Simulation of Railroad CTC Operations

culations are made at discrete time intervals, and the velocity, time, horsepower ex-
pended, braking, etc., are recorded at specified mileages. In addition, a velocity profile
is plotted. For the CTC simulation, which is only one of the many uses of this pro-
gram, condensed tables of running times are compiled for a range of trains of typical
characteristics. Using interpolation on these precalculated tables is both time and space
saving in the CTC simulation, although there may be a small loss of accuracy due to
rounding off siding-to-siding times to the nearest minute.

The Optimum Path Program

Because of limited storage in the computer, the simulation has been divided into
three separate programs, namely, an optimum path routine, the dispatching program
and a statistical analysis. A flow chart of the computer operation is shown in Fig. 1.
The optimum path routine assimulates the raw data and produces for each train a table
of the times it will enter and leave each siding and, in addition, indicates whether the
siding is adequate for the train, whether the power switch is facing or opposing the
train, and the type of task to be pertormed, if any. This optimum time line is the path
the train would take, allowing for schedules and tasks, but assuming there is no inter-
ference from other trains. The departure time from one siding and the arrival time at
the next specify the block requirements for each train. Since only one train may occupy
one block at a time, there is no need to record the distribution of time within a block.
The optimum time lines together with the track data form the input to the main
program.

The computer time for this routine averages 10 min, or approximately 10 percent
of that for the dispatching program. Incorporated in this routine are editing features
which detect most invalid data, and which from experience have proved very necessary.

The Dispatching Program

A block diagram of the dispatching program is shown in Fig. 2. Trains' optimum
time lines are read into the computer in the order trains are ready to enter the sub-
division. A horizon is set, determined by the earliest time that any train in the system
will reach its destination. All trains due to enter the system prior to this time are read
into the computer. By this means the number of trains being considered varies according
to the traffic density.

In a systematic manner, trains are moved forward as far as possible on a free
path. A train's progress is halted at the furthest adequate siding to which it can move
prior to a train ahead in the same direction, or to a conflict of time requirements for
a block. If progress is stopped by a conflict, the train number is stored in the appropri-
ate siding location, and the time is defined as the time the second train is to enter the
block. In this way all trains are moved as far as possible on a free path.

When all movements have been completed, the earliest of conflicts is found, together
with the trains involved. The CTC dispatcher normally plans meets and sets the switches
at sidings well ahead of the trains. Similarly, in the program the conflict is recognized
before either train enters the disputed block, and the decision is made at this time.
In the decision-logic phase of the program, the time-space around the two conflicting
trains is examined. For each train, information is collected on its priority, length, the
d3lay it will incur if the other proceeds, and for a scheduled train its slack time to the
next schedule point. The slack time is the difference between its minimum running time
and its schedule time. With the two basic alternatives of holding one train, the important
factors Influencing a decision are:

A Computer Simulation of Railroad CT C Operations 85

/TRACK DATA

( TRAIN DATA

RUNNING TIMES

OPTIMUM

PATH
ROUTINE

OPTIMUM
TIME LINES

DISPATCHING
ROUTINE

( DECISIONS

ORIGIN-
DESTINATION
PERFORMANCE

STATISTICAL
ROUTINE

STATISTICAL
ANALYSIS

Fig. 1 — The computer operation.

A Computer Simulation of Railroad CTC Operations

READ TRAINS UP
TO HORIZON

DETECT EARLIEST
CONFLICT. MOVE
ALL TRAINS PRIOR
TO CONFLICT

IS SITUATION CLEAR
FROM PREVIOUS CONFLICT

USE FORCE-ON
LOGIC TO CHOOSE
TRAIN TO MOVE

DECIDE TRAIN
TO BE HELD

IMPLEMENT
DECISION

RESET
HORIZON

Fig. 2— Flow chart of the dispatching program.

A Computer Simulation of Railroad CTC Ope rations

ARE TRAINS AT
ADJACENT SIDINGS?

IS AN INTERMEDIATE

SIDING ADEQUATE FOR

EITHER TRAIN?

INVESTIGATE

ADVANTAGES OF

SAWBY

FAVOURABLE

PLAN
SAWBY

UNFAVOURABLE

3 C

IS ONLY ONE OF THE
TRAINS ABLE TO
ACCEPT DELAY

IS ONLY ONE OF THE
TRAINS BLOCKED BY THE
THIRD

DELAY
THAT
TRAIN

DELAY
BLOCKED

TRAIN

USE PRIORITIES TO
WEIGHT DELAYS.
DELAY TRAIN WITH
LEAST RELATIVE DELAY

Fig. 3 — Decision logic (simplified).

1. the possible interference to other trains;

2. the ability of each train to accept a delay;

3. the priority and potential delay of each train.

For a meet, the skeleton logic is Bow-charted in Fig. I \ scheduled train can
"accept" a delay if it can be absorbed in it- slack time, i.e., il can suffei the delaj and
still maintain its schedule. In this connotation an unscheduled train can always "accept"
a delay. The disposition of other trains in the system Is checked t.. see whethei .i delay
to a third train would ensue if either train were allowed I" proceed li SO, the train
which would cause this delay is said to be blocked. In the situation ol tw> trains oppoi
ing one, the tendency is to delay the one When neither or both train> are blocked,
and neither or both can accept the respective delays, the priority and delay for each

88 A Computer Simulation of Railroad CTC Operations

train are compounded to choose the minimum weighted delay. Other factors affecting
a decision are the presence of one train in a terminal (since there is a premium on ter-
minal time), the use of grade sidings, and tasks to be performed at a siding.

The inclusion of short sidings in a configuration (or the running of long trains)
contributes to the complications of the decision logic. With short sidings, a saw-by
is sometimes a third alternative to delaying either train at its present siding. For this,
the long train must arrive first at a siding which is long enough only for the short train.
Since this decision causes delays to both trains, the saw-by logic measures this total
delay against a delay to either train, and if unfavorable returns to the decision logic
for a straightforward meet.

Nothing has yet been said about overtaking situations. Overtaking occurs mainly
while one train is performing a task, and hence little or no delay is caused to that train.
The use of sidings for overtaking is limited to those which are both adequate for the
train being passed and which can be entered through the power switch. The desirability
of an overtake is based not only on the relative speeds of the trains, but also on future
tasks, if any, and on scheduled arrival times.

When it has been decided which train is to be delayed, the opposing train is moved
to the siding location of the delayed train, and one of the two trains placed in the
siding, depending on the power switch disposition. If the delaying train has to run
through the passing track it is delayed an amount equal to the difference in time to run
through at normal speed and at siding speed. Its arrival and departure times at all
siding locations ahead are then modified by this amount. The clearance time of the
delayed train is then determined by the delaying train, and its future path modified
according^.

With the two trains still at the same siding, the program returns to the beginning,
the horizon is reset, more trains are brought into the system if necessary, and all trains
are mcved prior to conflict. If one of the two formerly conflicting trains is able to move
on a free path to an adequate siding at this stage, then the next conflict in time may
be accepted for resolution. However, if the situation does not clear itself, one of the
two trains must be forced to move on to a further sid'ng, leaving one train only at
each siding and thus available for arranging future meets. The situation occurs in times
of congestion and in particular when one train is longer than some sidings. Considerable
program development was necessary to handle the many conditions that occur in "force-
ons", which resu't mostly from saw-by meets. These situations occur in real operation,
for when a dispatcher makes such a meet, and allows a long train into sections of track
where the sidings are inadequate, he must accept the consequence of that meet by ac-
cepting further saw-bys or, if opposed by other long trains, by first forcing the train
on to an adequate siding and holding the opposing trains.

In this way the program is recycled until all trains have been processed. For each
decision made, a record is punched, and when a train reaches its destination the resolved
path of its arrival and departure times for each siding location is punched out.

The sequence of operation of the program is reviewed in time-distance graphs for
a specific example, in Fig. 4. On the vertical scale S^, S3 are 100 car sidings and S«.
60 cars. The horizontal scale is time. The manifest MF is 100 cars; the passenger P, 10
cars; and the wayfreight WF, 40 cars. Stage (a) represents the output of the optimum
path routine, that is. the path each train wou'd take neglecting the pressnee of the other
trains. In (b) the first three trains have entered the system, moved as far as possible
prior to the earliest conflict, and are at the siding locations indicated by the arrows.

A Computer Simulation of Railroa d CTC Operations 89

MF
(100) ^

l
I

\ I

-\T-

/\
/ \

±-4

p /

(10)4

1 /

/WF
/ (40)

(100)

(60)

S3
(100)

MF\

(100) \

/ \

/ \ /

\ 1 \ 1

* A

4 \ J \

s 1

(10)/

*1 /

r \

(o)

(b)

Fig. 4 — Phases of the dispatching program cycle.

Trains are halted at a siding prior to an event, hence tluir times an do) Identical. In
(c) the decision has been made to hold the manifi I Ml il 5 the pmcngfl P DU
been moved to 5! and the path of MF modified. The passenger P has readied the ter-
minal in stage (d) but the next decision must be made bcfori- the manifest movi
yond Si, as 53 is a short siding. The next operation would be to resolve thai conflict
between the manifest and the wayfreight.

90 A Computer Simulation of Railroad CTC Operations

I, I'. [AY ''".|.

I i

► lit scr. i«i pfKiur
pc ■ t « ? :■

Fig. 5 — Simulated train chart.

The program uses the minimum clock technique, moving from conflict to conflict.
Based on the activity of an average subdivision, this represents a considerable saving in
time over the master clock method. The highest clock time is 10,000 min (just 40 min
short of 1 week) and the smallest interval is 1 min. For an average subdivision of 15
sidings and 120 trains per week, the dispatching program takes about V/z hr, which
gives a simulation ratio of 100 to 1. Since the relationship between computer time and
train density is non-linear, this ratio will decrease rapidly with higher density traffic.
Because of the limitation of 4000 words of storage in the IBM 650, the number of
sidings cannot exceed 38 and the number of trains in the system at any one time is
restricted to 10. The program itself occupies 3000 locations in memory.

The Train Charts and Analysis Program

The punched cards of the trains' resolved paths are sorted and listed off-line on an
IBM 407 to produce scaled time-distance charts of the simulated dispatching. Each train
is allocated a digit, and a point is plotted for the time a train arrives at, and leaves a
siding location, as shown in Fig. 5. The day is divided into 3 periods of 8 hr. The ends
of the sidings are considered to be 1 mile apart for plotting purposes. Where only one
mileage appears against a siding number it is either a junction or an electric lock siding.
The dotted lines show the difference in time between the actual arrival time and the

A Computer Simulation of Railroad CTC Operations 91

scheduled arrival time. For unscheduled trains the equivalent of the scheduled time is
the time the train would have arrived without any interference

The statistical program utilizes the resolved origin-destination performance, and
the decision and delay data from the dispatching program to prepare an analysis of the
siding configuration and the traffic pattern imposed. The most general indication of
performance is given by the total train interference in train-minutes. This interference
is also given as a distribution over various trains or classes of trains. Some of the
other statistics derived include:

1. The lateness of scheduled trains leaving and arriving at terminals

2. A distribution of the number and duration of delays by sidings.

3. The accumulated number of meets and passes occurring during the same hour
of each day for the period. (This shows the traffic peaks).

4. The average delay per meet or pass.

5. A distribution of the activity of the system. This is measured by the number
of hours in which no meets, one meet, two meets, etc., occur.

6. The distribution of interference to each class of train caused by each class ol
train (e.g., to passenger trains by manifest trains).

C. PROJECT EVALUATION AND DEVELOPMENT

Evaluation of Results

When an analysis of a subdivision is required, about five or six configurations are
initially chosen for comparison. These configurations, which range from an expected
minimum to maximum number of sidings, are based on proposals from local and Re-
gional operating personnel, and the Headquarters Transportation Department. Because,
in most cases, CTC is being installed to supersede train-order operation on a subdivision,
and the cost of building complete new sidings is high, the existing siding locations impose
many constraints on the location of sidings to be signalled. On train-order subdivisions
the siding spacing is normally much less than that considered necessary for CTC opera-
tion. Hence the problem often reduces to deciding which of the existing sidings should
be signalled, and in some cases extended.

From the statistical analysis and from the train charts, the Transportation Depart-
ment can very quickly judge the feasibility of a configuration. The charts highlight any
extraordinary situations, or poor dispatching, which may be hidden in a statistical
summary. The results of'.en lead to further proposed configurations, which are simulated
before the economics of the different configurations are studied,

It is expected that the introduction of CTC will reduce both the operating
of a subdivision and the cost of interference to traffic on thai subdivision. The main
items contributing to the operating costs, which are significantly different in train order
and CTC operation, are operators' wages, maintenance ot signals and maintenance of
way. Operators' wages, and to some extent the maintenance costs, are independent ol
the CTC configuration. To the railway the main measurable costs ol Interfereno
car time and locomotive time. The potential reduction in costs must be weighed against
the capital cost of CTC (or the annual interest and depredation cost) Trends of the
capital cost of CTC and the resulting interference for a typical subdivision are shown
in Fig. 6.

In addition to the reduction of interference costs, there an benefits ol CTC which
have as yet only been measured qualitatively The most apparent of these are Imp

92 A Computer Simulation of Railroad CTC Operations

2000 2500 3000 3500

INTERFERENCE (TRAIN MINUTES)

Fig. 6 — Capital cost and interference.

operating safety, faster and more consistent service to both freight and passenger cus-
tomers. In a ccmplete appraisal of a configuration the quantitative value of these factors
must be included.

In most of the subdivisions studied to date, there is no proposed configuration
giving acceptable performance, which stands out clearly as the economic optimum.
Usually, within the limits of permissible performance there is a steady improvement in
performance with increased total cost. Two problems, however, can be easily solved:
if the capital cost is fixed, the configuration giving the best performance can be found,
or if the desired performance is specified, the cheapest configuration to attain this can
be found. In general, considerations other than performance limit the choice of con-
figuration. Since the studies are based on current traffic levels it is necessary in design-
ing the CTC configuration to provide enough flexibility to handle the expected traffic
five to ten years hence. It is now a standard practice in determining the most suitable
configuration, to choose one which can be easily and efficiently expanded for greater
capacity.

Program Development

The current simulation program produces a redispatch reasonably close to that of
a CTC dispatcher. It is to be stressed that the program is not designed to operate on
line and in real time to do the dispatcher's job. Its main function is in p'anning require-
ments for CTC installations and has been used for this purpose by the Headquarters
Transportation and Maintenance Department since July 1960. It has also proved valu-
able in planning siding extensions and retirements on existing CTC subdivisions. Fig. 7

A Computer Simulation of Railroad CTC Operation:

► III SO0. 0>T M«I0

I.I I

II.} I

h.i i
m.i •

tl.?

• •.0

11.0

H.J

»».»

tl.t
is.*

IS.'*

tt.l

II
it

tl.t
It.*

II
II

II. 1

rt.i

ii.i

n.«

•s.i

• t.r
ir.t

101. I II

lot. I ir

IC5.S If

Fig. 7 — Simulated and actual dispatching.

94 A Computer Simulation of Railroad CTC Operations

is a comparison of simulated and actual dispatching on an existing CTC subdivision.
The simulated dispatching is shown by a solid line. Where the actual dispatching is
different from the simulated, it is shown by a dotted line. Overall, the differences are
minor, and the simulation compares favorably with the CTC dispatcher.

The current program has certain limitations, some of which are due to memory
size of the IBM 650. At present, for example, only one subdivision, or a section of
track with not more than 38 sidings, can be simulated. A new program, designed pri-
marily for the same purpose, is now being developed for use on an IBM 7070. This will
enable a greater length of track to be simulated, and will inccrporate more extensive
decision logic. Since the program was first used, much experience has been gained of
CTC operation and of subdivisional problems, and this knowledge is a valuable back-
ground for future programs.

A more general simulation of railroad operations is also planned. In this it is
intended to simulate traffic over a large network, such as Montreal and Toronto to
Winnipeg, and to incorporate in the simulation the operation of major yards.

With the availability of a general and a detailed simulation program, many oper-
ating problems can be studied. The detailed program will continue to be used for CTC
subdivision studies. In addition it may help in an understanding of the theory of single
track traffic flow by studying the relationships between interference, traffic density, sid-
ing spacing, train priority and velocity. The general simulation could be applied to such
problems as the scheduling of traffic between major terminals, and the operating effect
of varying the number and length of trains.

We now have a technique to compare CTC configurations quantitatively in terms
of performance and interference. The effect of these factors is, however, not sufficiently
understood to apply cost-effectiveness measures to the quantitative results. This is a
worthwhile area for intensive research.

Average Tie Life — An Interpretation*

By C. J. CODE
Assistant Chief Engineer — Staff, Pennsylvania Railroad

The cross tie statistics for Class I Railroads which appear annually in the Bulletins
of the American Railway Engineering Association and in the Cross Tie Bulletin are of
great interest to many people in the railrcad field. This interest extends from tcp man-
agement down into the ranks of the engineering and purchasing departments and has
not infrequently been the subject of editorial comment in railway magazines.

That these figures are subject to misinterpretation and misuse probably does not
distinguish them materially from other types of statistical information. The first cause
of misinterpretation is the tendency to look at the figures for one years renewals, or
even the average figures for five years, and place undue weight on the results as reflect-
ing differences in maintenance efficiency or economy — for instance, as between two rail-
roads. Again, if we are to evaluate the benefit of any change in practice which is ex-
pected to produce increased tie life, we must have an accurate idea of present-day tie
life on which to base a comparison. I think most of you will agree that the five-year
average including 1961. which could be interpreted as reflecting an average life of 62
years, is deceptive to say the least, and that these figures can scarcely be taken as truly
representative of average tie life.

Actually, to get a true overall picture of tie life, we should have an average over a
sufficient number of years to include the fife of the ties which remain in track longest.
This would mean, in my opinion, an average over a period of 50 to 60 years. However,
we have not been using creosoted ties for a major percentage of renewals for 60 years.
and it is necessary in obtaining a realistic average to confine our study to a period
during which the present-day standard of creosoted ties represents all but a very small
percentage of the ties used. It is also desirable if not essential to choose years for the
beginning and end of the period such that the average condition of ties in track is about
the same at the beginning and end of the period. I have chosen for these reasons the
period from 1934 to 1957, incl.

In interpreting the statistics, it is also necessary to take account of the fact that
mechanical wear and damage has an important effect on tie life and that consequently
the average life in a heavy-tonnage track in territory where there are many sharp curves
heavily elevated is going to be quite different from what it is under light traffic at low
speed on tangent and light curves. One cannot take the overall average, apply it to
any specific situation, and say that the ties in that location are expected to equal the
overall average life. For instance, if I make a study of the economic benefit of a certain
change in tie practice which will apply only to main track, or to main track in a specific
territory, and base the study on an average life which includes 40 percent of side and
yard tracks, I am only deceivim: myself.

How this can work out in a specific instance is shown on Table I. which repre-
sents a combination of two railroads, one eastern and one western, for which tin- statis-
tics were readily available, and in which the average renewals over the 24-year period,
1934 to 1957, inch, are broken down by classes of track. In making this breakdown
of mileage I have had to make some assumption-, but these are not such as to distort
the overall picture.

"Presented before the annual convention "f the Railw.iy Tie Association, Minneapolis, Minn,
October 24, 1962.

96 Average Tie Life — An Interpretation

Table I
Railroads A and B
Tie Life - 24 Year Average
1934 to 1957 Incl.

Class of Track and Miles on Ties in Ave. Ties Renewed

Annual Gross Ton Miles Cros3 Ties Ties/Mile Track Life Ties/Mile/Year Per Year Total

2560 19,700,000 60

3070 18,650,000 60

2710 3,210,000 50

2970 6,870,000 35

3100 10,400,000 30

3240 38,100,000 32

3100 10,050,000 25

3240 6,480,000 25

3250 6,630,000 20

39,657 120,090,000 35 86.5 3,428,000

(a) and (b) reflect differences in ties per mile.

The values of average life used for various classes of main track are based on my
own studies over the past 17 or 18 years in which I have had the opportunity to follow
the performance of dated test ties in detail throughout their life to date. This has been
augmented by observation of older ties of known vintage based on the type of end irons
used, and on overall observation of tie performance.

You will note that average life of ties in main track is shown as varying from 20
to 35 years for all but the lightest traffic lines, and these figures are pretty well sub-
stantiated as far as I am concerned by the observations referred to above. The 20-year
average for main-line heavy traffic could be further broken down to show some loca-
tions where life is only 15 years. The figure of 50 years for light traffic branch lines
and the figure of 60 years for side and yard tracks, are pretty much assumptions, but
they are assumptions which are necessary in order for the table to balance out to an
overall average of 35 years for the two railroads, which figure is obtained from the
published tie statistics for the 24 year period.

Since first preparing the table I have had a look at some 40-year-old ties in little
used side tracks, and I must confess that based on this observation, my 60-year figure
for yard and side tracks looks too optimistic. Nevertheless, I am quite hesitant to make

Side and Yard Tracks

(a) 7697

(b) 6080

Main Tracks
Less than 2 Million
Gros3 Ton Miles

1185

2 to 5 Million
Gross Ton Miles

2310

5 to 10 Million
Gross Ton Miles

(a) 3355

5 to 10 Million
Gross Ton Miles

(b)ll,750

10 to 25 Million
Gross Ton Miles

(a) 3240

10 to 25 Million
Gross Ton Miles

(b) 2000

Over 25 Million
Gross Ton Miles

2040

42.7

328,000

51.2

311,000

54.2

64,000

84.8

196,000

103.3

347,000

101.3

1,190,000

124.0

402,000

129.5

259,000

162.5

331,000

Average Tie Life — An Interpretation

'-7

Table II
Railroads A and B
Tie Life - 24 Year Adjusted Average
1934 to 1957 incl.

Class of Track and Miles on
Annual Gross Ton Miles Cross Ties

Side and Yard Tracks (a) 7697

(b) 6080

Main Track
Less than 2 Million
Gross Ton Miles

2 to 5 Million
Gross Ton Miles

5 to 10 Million
Gross Ton Miles

10 to 25 Million
Gross Ton Miles

Over 25 Million
Gross Ton Miles

1185
2310

(a) 3355

(b) 11,750

(a) 3240

(b) 2000
2040

Total Ties

Ties/Mile in Track

2560 19,700,000
3070 18,650,000

2710 3,210,000

2970 6,870,000

3100 10,400,000

3240 38,100,000

3100 10,050,000

3240 6,480,000

3250 6,630,000

Ave. Ties Renewed

Life Ties/Mile/Year Per Year Total

51.2

394,000

61.4

373,000

54.2

64,000

84.8

196,000

103.3

347,000

101.3

1,190,000

124.0

402,000

129.5

259,000

162.5

331,000

39,657 120,090,000 33-8* 89.7* 3,556,000*

(a) and (b) reflect differences in ties per mile.
* This is an adjusted figure - see text.

the upward revision of about 5 percent in the life of main-track ties which is necessarj
to make the table balance with, for example. 50 years for side-track life.

It is quite likely that in the 24-year period covered by the average, we have not
experienced the renewal of a sufficient percentage of our side-track tie- to have thl

age figure truly representative of the life of these ties, li I assume SO year Bfe for tie-
in yard and side track, and make do change in tnj figures for main track, I arri<
an overall average life of 33.8 year- instead of IS years It seems t" me <i"'11' likely
that this is a more realistic overall average life than the ; years shown in Table I
I have prepared Table II to show this situation.

I am sure there are railroad- where there i- not such B wide variation in main

track tie life as I have shown, and perhaps a single figure would mi main

track mileage on such a road; however, I am sure that an adjustment should be made
to allow for the greater life in yards and side tra< I
has a single main line with 408 miles ol track maintained on
main track and L9S side and yard tracks, with an
year, representing an overall tie life oi

98 Average Tie Life — An Interpretation

and side tracks with 2840 ties per mile in such tracks vs. 3250 ties per mile in main
track, we arrive at an average life for main track of 25. 2 years, which is probably not
far off for this road.

Obviously, maintenance policy, in both good times and bad, has a marked effect
on tie life. A policy of careful economy disregarding refinement of line and surface
will produce longer-than-average life, while a policy of "nothing but the best is good
enough" will produce a shorter-than-average life. A change in maintenance policy can
easily make a difference of 5 or 10 percent in apparent average tie life.

So that you do not have to take on faith alone my figures for tie life in main
track, I am including a set of photographs which illustrate the appearance of ties in
main track after 3 years, 7 years, 10 years, 13 years, 15 years and 16 years in heavy
traffic main line, after 25 years in a branch line of moderate traffic, and after 32 years
in a track having frequent low-speed movement of light passenger trains, but no freight.
You will note that some have already been removed from track after 13 years. At 15
and 16 years many are ready for renewal. A recent check of about 1000 ties installed
15 years ago in a series of 6-deg curves in a track carrying an average of 60 million
gross tons annually, shows an estimated life, based on the percentage remaining in track,
of 14 years with one type of anti-splitting device and 16 years with another.

In the lighter traffic situation after 25 and 30 years, heavy renewal is indicated,
and was in fact in progress or about to start when I took my pictures.

In many cases these pictures show the worst ties of a given lot; however, I have
tried to include also some pictures of the average or better ties.

I have found that under heavy traffic most ties still look new after five years. At
from 8 to 10 years they start to show deterioration. At 13 to 16 years many are ready
for renewal, although many still have several years life.

The appearance of these ties, and of the ties associated with them, leads me to
believe that the table presented gives a reasonable interpretation of tie life under the
conditions with which I am familiar. A similar table can be constructed for any rail-
road or group of railroads by collecting certain basic information.

Tie statistics must be averaged over a long period of time to be meaningful, and
must be interpreted in the light of knowledge of the variations of life with different
classes of track and service conditions.

Average Tie Life — An I nterpretation

New ties, recently installed.

Hull. 573

100

Average Tie Life — An Interpretation

3 year old ties, combined seasoning and treatment.

7 year old ties.

A v e r a ge Tie Life — An Interpretation

101

10 year old ties.

102

Average Tie Life — A n Interpretation

10 year old ties.

Average Tie Life — An Interpretation

103

13 year old ti<

104

A v e r a ge Tie Life — An Interpretation

15 year old ties.

Average Tie Li f e — An Interpretation

105

16 year old ties.

106

Average Tie Life — An Interpretation

■■

16 year old ties.

Average Tie Life — An Interpretation

107

16 year old ties.

108

Average Tie Life — An Interpretation

ysfis*1

16 year old ties.

Average Tie Life — An Interpreta t i o n

10"

25 year old ties — medium-traffic branch line.

Average Tie Life — An Interpretation

32 year old ties— light-tonnage, slow-speed suburban passenger traffic.

Cut Maintenance Costs With
Dependable Aeroquip Hose Lines

Aeroquip Flexible Hose and Reusable
Fittings are designed to cut costs two ways:
Dependable replacement lines can be
assembled quickly, easily in your shop as
they are needed. There's no costly in-
ventory of factory-made lines; a few coils

of bulk hose and some reusable fittings
meet every requirement. Second, you save
on future replacements because every
Aeroquip Fitting is detachable and com-
pletely reusable. Call your Aeroquip Dis-
tributor today!

AEROQUIP HOSE LINES FOR RAILROADS

For Hot Water, Fuel and Lube
Oils, use Aeroquip 1534 Hose and
Fittings. Sizes from Va" to 2Vi " meet
all locomotive and equipment require-
ments. Special high temperature hose
types also available.

For Air Brake and Signal Lines,

use Aeroquip 1531 Air Hose and
Fittings. Hose has internal reinforce-
ment for longer service life, lower
initial cost.

eroquip

AEROQUIP CORPORATION • JACKSON, MICHIGAN
INDUSTRIAL DIVISION

INDUSTRIAL PLANTS: VAN WERT, O.J BURBANK, CALIF.; DALLAS, TEX.;

PORTLAND, ORE.; CRANBURY, N.J.; ATLANTA, GA.

In Canada: Aeroquip (Canada) Ltd., Toronto 19, Ontario

In Germany: Aeroquip GmbH., Baden Baden-Oo>

AEROQUIP PRODUCTS ARE PROTECTED BY PATENTS IN U.S.A., CANADA AND ABROAD
FLEXMASTER is an Aeroquip Trademark

AEROQUIP PRODUCTS ARE DESIGNED FOR BETTER RAILROADING

FLEXMASTER Pipe Coupling Diesel Manifold Clamp

Aeroquip Automatic Fueling Unit

£-x-t-e-n-d i-l-Q JZ-i-ft-el

USE TIE PLATE
LOCK SPIKES

One-piece Design

LOCK SPIKES hold tie plates firmly in place on
cross-ties and bridge timbers.

LOCK SPIKES are quickly and easily driven,
or removed, with standard track tools.

Driven to refusal, the spread shank is com-
pressed by the walls of the hole. Tie plates are held
against horizontal and vertical movement under
spring pressure. Play between the spike and the
hole is eliminated — abrasion and seating of tie
plates is overcome.

LOCK SPIKES hold their position in the tie,
and redriving to tighten the plate is not required.
They provide a quiet and strengthened track.

Annual cost of ties and maintenance expense is
reduced by extending the life of ties and holding
gage. Here is one answer to conservation of ma-
terials and labor. Write for free folder.

BERNUTH, LEMBCKE CO., INC.

420 Lexington Avenue, New York 17, N. Y.

Actual
Size

6RS Announces ...ROLKODE

A New Communication System SJJJT"

SOLID
UNITS

for centralized

traffic control

for supervisory

control systems

Advantages of the System:

• ROLKODE is up to 50%
faster than time coded sys-
tems.

• When combined with other
services, ROLKODE sim-
plifies line circuit require-
ments.

• ROLKODE operates on line
wire, cable, microwave, or
telephone company facili-
ties.

• ROLKODE eliminates
need for line battery or
stand-by a-c.

Control office solid state counter.

Typical coding unit.

ROLKODE is simple, fast, and
economical. Its advanced electronic
circuitry, solid Btate units, new and
improved relays, and unusual adapt •

ability will make your oexl
installation the most efficient avail-
able today.

FOR MORE INFORMATION ASK FOR PUBLICA

GENERAL RAILWAY SIGNAL COMPANY

ROCHESTER 2, NEW YORK NEW YORK 17, NEW YORK CHICAGO 1, ILLINOIS ST. 10UIS 1. MISSOUH

The custom-built assembly shown
above and to the right is an all-purpose
rig designed to give maximum flexibility
in coating and painting work. It was
designed for field application of paints,
lacquers, vinyls, cutback asphalts,
creosotes, heavy oils and greases.
<- It uses the economical Graco Hydra-
Spray Process, and proves once again,
you get the job done faster and better
with Graco than with any other coat-
ing system.

If speed of coating application, and
material savings are important to you,
write today for all the details of the
Graco Hydra-Spray Process.

FREE!

Graco Engineers are prepared to help you in the design of your
paint and material spray assemblies. Your Graco Railway Rep-
resentative will be glad to explain the many benefits of this
service. Write or call him . . . today!

Graco

GRAY COMPANY, INC.

MINNEAPOLIS 13, MINNESOTA

RAILWAY DEPARTMENT

JOHN P. McADAMS, Eastern Sales Representative
2304 Wilson Boulevard, Arlington, Virginia

CHICAGO— (Broadview, III.)

R. D. Worley

3030 South 25th Ave.

CLEVELAND

M. H. Frank Company, Inc.
1202 Marshall Building

HOUSTON

Houston Railroad Supply Co.
1610 Dumble Street

PHILADELPHIA

The A. R. Kidd Co.

1036 Suburban Station Bldg.

LOUISVILLE

T. F. & H. H. Going
6308 Limewood Circle

ST. LOUIS

The Carriers Supply Company
818 Olive Street

NEW YORK— Newark, New Jersey

R. A. Corley
744 Broad Street

SAN FRANCISCO

The Barnes Supply Company
Rm 504, 74 Montgomery Street

TWIN CITIES— SI. Paul, Minn.

The Daniel L. O'Brien Supply Compare
Endicott-On-Fourth Bldg,

WASHINGTON— Arlington, Va.

Southeastern Railway Supply, Inc.
2304 Wilson Blvd.

MONTREAL— Ontario, Canada

International Equipment Co., Ltd.
360 St. James Street West

the

JACKSON UTILITY TAMPER

An inexpensive, easy-to-maintain Multi-purpose Tamping Machine!

Here's a tamper involving very moderate investment that is gaining tremendous
favor with trackmen everywhere. It can be used to great advantage for jack-
tamping, spot-tamping and production tamping in moderate to high raises ... a
truly multi-purpose machine that can be kept working virtually all the time . . . and
one that returns one of the biggest dividends of any investment in track equipment.

PERFECT FOR SPOT TAMPING in all ballasts and
all conditions. Power options are available to suit in-
dividual requirements.

AS A JACK TAMPER-lts fast, accurate
and positive in reaching and holding the track
raise and level. It easily keeps ahead of large
production tampers and is adapted to existing
surfacing devices.

AN EXCELLENT PRODUCTION TAMPER

—Produces complete consolidation of ballast
in out-of-face raises with maximum stability
right under the rail— the load bearing zone.

SIMPLE CONTROLS

Fast, easy to coordinate.

Manual-hydraulic.

Let us furnish you with complete
information concerning this excep-
tionally advantageous machine.

As always, direct sales,
leases and service to all
U.S. railroads.

JACKSON

VIBRATORS. —

LUOINGTON. MICHIGAN. US»

Machine knocks off, ejects ties, lines track behind plow while above subgrade.

Mannix AUTO-TRACK

liminates 15 to 20 Men

PLOWING, TIE REPLACEMENT, TRACK ALINEMENT are
accomplished faster and with fewer men using the new
Mannix AUTO-TRACK unit.

Hydraulic hammer on each side knocks tie down. Con-
veyor belt ejects tie to either side. ^

NOW, lease or purchase MANNIX Auto-Track
Equipment to operate with railroad crew for
greatest convenience and maximum savings.
Write for details. Arrange showing of operating
films with no obligation,

^,™ 'mm* a

to • ^arn <*
• ♦ * k_J

Lining head controlled by levers
near rear end where operator
observes alinement. One man
replaces lining crew.

MANNIX

INTERNATIONAL

Phone: Liberty 5-0411

Box 7485

MINNEAPOLIS 22, MINNESOTA

your

service

for

all types of cranes

d iesel wreckers

pile drivers

buckets

ORTON

CRANE & SHOVEL CO.

608 S. DEARBORN ST.
CHICAGO 5, ILLINOIS

DANIEL A. COVELLI

President

Representative* in Principal Cities

with America's most complete line of
special trackwork: For Railroads,
Mines and Industries — A complete
line of frogs, switches and crossings •
Trackwork for installation in paved
areas • Manganese steel guard rails
• Automatic switch stands • Samson
switch points • Snow-Blowers • Switch
point guards • Rail and flange lubri-
cators • Tie pads • Racor studs • Dual
spike setters • Dual spike drivers •
Car retarders.

with America's most complete track-
work manufacturing facilities: Coast
to coast to serve your needs.

»► RACOR PLANTS:

Buffalo • Chicago Heights • Los Angeles •
Niagara Falls, Ontario, Canada • Pueblo

• Superior.

• RACOR SALES AND ENGINEERING

Chicago • Cleveland • Denver • Houston •
Los Angeles • Louisville • New York •
Philadelphia • Richmond • San Francisco
St. Louis ■ St. Paul. In Canada: Dominion
Brake Shoe Co., Ltd., Montreal, Quebec

• Niagara Falls, Ontario • Winnipeg,
Manitoba • Vancouver, B. C.

* RACOR RESEARCH:

Chicago • Mahwah, N. J.

with America's most complete track-
work engineering service: This lies in
making available to our customers
Racor's engineering experience —
practical experience from years of
designing and manufacturing . . .
advanced experience solving tomor-
row's trackwork problems today in
Racor research laboratories.

Why not let us help you with your
trackwork problems?

RAILROAD PRODUCTS DIVISION

-<JS2>" 530 Fifth Avenue, New York 36, New York

'[BrakeShoe,

CHANNELOC

...THE

HARD-WORKING

RAIL ANCHOR

■

S'^7 ■ ,

39 \ ••■. -

HP1'

Channeloc's construction gives it the brute strength
to grip the rail with extra force — with plenty of
reserve strength for reapplication. And the way we
make it gives it the temper, the spring, to fit itself
to variations in the edge thickness of rail bases . . .
no shims. This is the anchor we introduced just
two years ago, and millions have already been
ordered by a large number of Class I roads.

ml RUE I EM PER.

RAILWAY APPLIANCE DIVISION
1623 EUCLID AVE.. CLEVELAND 15. OHIO

P. O. Box 10378

LOgan 6-7922

GREENHEART, INC.

1431 N. E. 26th Street
FORT LAUDERDALE, FLORIDA

President — John L. McEwen — Quarter Century Experience

IMPORTERS:

Greenheart Piles, Lumber, Timbers Long Length
MORA EXCELSA— Lumber and Timbers

Teak and other Woods from Burma, Siam, Australia,
Africa and South America

the world's largest
selling rail anchor

876R

IMPROVED
FAIR

RAIL ANCHOR

POWER

DESIGN MEANS

• longer service life
/ • Greater holding power
f^Faster, easier installation

fHIP.*M.22:

Division of Poor and Company

CHICAGO. NEW YORK, DENVER, ST. LOUIS, BOSTON, ST. PAUL, WASHINGTON. SAN FRANCISCO, MEXICO CITY

mmmgm

Assure lower maintenance costs,
better performance with...

;:::::::;:::::::x:>-::::: :■::-: ::::-::: ::::.::- 'xVy::':'::: yXyMVy*

TEXACO

Railroad Lubricants

and Systematic

Engineering Service

TEXACO inc.

RAILWAY SALES DIVISION

135 East 42nd St., New York 17. N. Y.

NEW YORK • CHICAGO • SAN FRANCISCO • ST. LOUIS • ST. PAUL . ATLANTA

Hubbard Super Service Alloy Spring Washers

Hubbard Super Steel Alloy Spring Washers

Hubbard Track Tools

Hubbard Tool Division

UNIT RAIL ANCHOR CORPORATION

New York Pittsburgh Chicago

Unit Rail Anchor

UNIT RAIL ANCHOR DIVISION

UNIT RAIL ANCHOR CORPORATION
NEW YORK PITTSBURGH CHICAGO

FIBRE REPAIR PARTS

for
INSULATED JOINTS

The Effectiveness of the Joint

Is Dependent Upon the Quality,

Workmanship and Fit of Every Part.

Rail Joint Company

Division of Poor & Company, (Inc.)
New York 7, N. Y.

Model N U Tie Cutter

HERE IS THE WINNING TEAM

The Woolery NU Tie Cutter and the Woolery Tie-end Remover preserve the line and surface
of the track and at the same time reduce the cost of tie renewals. Ties can be removed
without trenching, jacking up track or adzing tops of rail-cut ties. With this team you simply
cut both ends of tie, pry out center piece, insert in its place the tie-end remover and out
go the tie ends pushed by the double acting, double ended hydraulic cylinder of the Tie-
end remover.

FOR HIGHEST EFFICIENCY USE TWO TIE CUTTERS WITH ONE TIE-END REMOVER

WOOLERY MACHINE COMPANY

MINNEAPOLIS, MINN

OCV WOODINGS-VERONA TOOL WORKS

\f Pioneer Manufacturers

oi
HIGH GRADE TRACK TOOLS

and
SPRING WASHERS FOR TRACK

Since 1873

VERONA. PA. CHICAGO. ILL.

WOODINGS FORGE 6c TOOL COMPANY

Makers

WOODINGS RAIL ANCHORS

VERONA. PA.

CHICAGO.

ELL.

Notes on

Railroad Location and Construction Procedures
from the School of Experience

By J. A. Given

A series of notes, comments, short-cut methods and "tricks of the
trade" written by a railroad location engineer of many years of
practical experience for the benefit of young engineers.

Price $0.50

AMERICAN RAILWAY ENGINEERING ASSOCIATION

59 East Van Buren Street

Chicago 5, III.

VEGETATION CONTROL

CHEMICALS

READE MANUFACTURING COMPANY, INC.

Jersey City — Chicago — Minneapolis — Kansas
City — Birmingham — Stockton

SERVING RAILROADS OF AMERICA FOR
MORE THAN FORTY YEARS

E
D

A
N
D

S
H

C
O
N
T
R
O
L

Here are the up-to-date facts on the SPENO Ballast
Cleaning and the SPENO Rail Grinding Services.

BALLAST CLEANING

SPENO Engineering and Research has de-
veloped a superior screening arrangement so
that we are now using an improved Ballast
Cleaner with greater efficiency.

RAIL GRINDING

Our Rail Grinding Service has been so well
received we are now building a THIRD Rai
Grinding Train to take care of the increased
demand.

SPENO is constantly developing means jot
better service to make sure that the Railroads
receive everything they pay for — and more

(yfoz?~/z&& tfie ^acfomds yn^~naae used; as.

FRANK SPENO RAILROAD BALLAST CLEANING CO., INC.

COT'

IIII'.IHl

Clark Sir
East Syracusf

THE TRASCO
AUTONOMIC CAR RETARDER

CLAMPS IN PLACE
ANYWHERE IN TRACK

SIMPLE — EFFECTIVE — INEXPENSIVE

TRACK SPECIALTIES CO.

GENERAL MOTORS BLDG.
NEW YORK 19, N. Y.

American Railway

Engineering Association— Bulletin

Vol. 64, No. 574 November 1962

REPORTS OF COMMITTEES

1 6 — Economics of Railway Location and Operation 111

9 — Highways 131

1 3 — Water, Oil and Sanitation Services 139

1 4 — Yards and Terminals 159

20 — Contract Forms 187

25 — Waterways and Harbors 197

6 — Buildings 213

The reports in this issue of the Bulletin will be presented to the 1963 Busi-
ness Meeting of the Association at the Conrad Hilton Hotel, Chicago, March
15-16. Comments and discussion with respect to any of the reports are solicited,
and should be addressed to the chairman of the committee involved, in writing
in advance of the Meeting, or from the floor during the Meeting.

BOARD OF DIRECTION
1962-1963

President
C. J. Code, Assistant Chief Engineer — Staff, Pennsylvania Railroad, Philadelphia 4, Pa.