Railway Demand Forecasting and Service Planning Processes: Prepared For: Rail Freight Service Review
Railway Demand Forecasting and Service Planning Processes: Prepared For: Rail Freight Service Review
March 2010
March 2010
Table of Contents
1. Purpose of the Project ................................................................................................................................. 4
2 | QGI Consulting
Railway Demand Forecasting and Service Planning Processes
March 2010
QGI Consulting |3
Railway Demand Forecasting and Service Planning Processes
March 2010
1. Purpose of the Project
On May 30, 2007, the Government of Canada announced the introduction of a Bill, which contained improvements to
the shipper protection provisions of the Canada Transportation Act. The announcement also indicated that the
government would undertake a review of rail freight service.
The overall objectives of this review are to:
• Conduct a review of the rail‐based logistics chain (including shippers, terminal operators, ports, and
vessels), with a focus on service provided to Canadian shippers and customers by Canadian National
Railways (CN) and Canadian Pacific Railway (CP) within Canada, including to and from ports and border
crossings;
• Identify problems and issues with respect to railway service including those stemming from other
elements of the logistics chain;
• For shippers located on shortlines and experiencing problems with rail service, examine the relationship
between shortlines and the main line carriers to determine whether such problems are attributable to
service, operating, or marketing practices of the main line carriers;
• Identify best practices and how these can be expanded to address service issues; and
• Make recommendations on how to address these problems and issues, including both commercial and,
if necessary, regulatory solutions.
The review is being conducted in two stages. The first stage consists of quantitative and analytical work. In the
second stage, draft recommendations are being developed by a Panel of three eminent persons based on the results
of the analytical phase and any other relevant information that is available. The Panel will consult stakeholders on the
draft recommendations and submit a final report to the Minister of Transport, Infrastructure and Communities.
On July 24, 2008 Transport Canada released the terms of reference for the quantitative and analytical stage of the rail
freight service review. This report has been prepared in response to the requirements of the Request for Proposals
(RFP) covering the description of railway demand forecasting and service planning processes. The objective of this
report is to provide a comprehensive description of how railways forecast demand, how and to what extent railways
interface with their customers in developing their demand projections, and how demand projections are transformed
into railway service and asset plans. This report also describes the railways’ ongoing short term planning and
management activities as they manage the asset and service plans developed through these annual planning
processes.
4 | QGI Consulting
Railway Demand Forecasting and Service Planning Processes
March 2010
2. General Approach
The railway planning processes described in this report are based on extensive interviews with CN and CP subject
matter experts directly involved in and responsible for the development of the railways’ demand forecasting and
service planning processes. QGI has also relied on the experience and background of the team members that
participated in the interviews and prepared this report.
Prior to conducting the interviews QGI provided CN and CP with a discussion document that identified the specific
topics to be discussed for each element of the planning process. Key topics of discussion included the following:
Demand Forecasting
1. Medium to Long term planning process (annual and 3‐5 years out)
a. How customer demand information is collected and used in the broader planning process
i. Key challenges associated with collecting and inputting customer demand projections into overall
process
ii. Differences in customer/market inputs across business lines and by shipper size
iii. Systems used to capture and analyze demand
iv. Demand forecasting units used – e.g. tons, cars, containers, etc.
b. Key “non‐customer” factors accounted for in demand planning and process for balancing between
customer stated demand and what ends up in the final plan
2. Short term demand forecasting and plan management (monthly and quarterly)
a. Periodic demand adjustments
i. Key factors that drive changes to internal demand forecast planning during the course of a year
ii. Frequency of adjustments
iii. Nature of ongoing customer dialogue with respect to shifts in demand due to: seasonal factors,
changes in market conditions (demand surge or reduction), changes resulting from disruptions
iv. Process for communication of short term changes in demand to operating departments
b. Forecast Performance Measurement
i. Nature and frequency of existing measurements
ii. Mechanisms for providing customers with feedback on accuracy of their projected demand
c. Demand forecasting outputs
i. Structure of outputs (major flows (corridors), origin‐destination demand forecasts, etc)
ii. Internal users of demand forecast information and for what purpose
d.Key railway challenges in development of short and long term demand forecasts
Network and Asset Planning Issues
1. How Marketing demand forecast is used for asset planning for: train service, crews, locomotives, rail car fleets and
track capacity
2. For each/all of the above asset planning processes:
a. Definition of planning cycle
b. Systems or models used
c. Processes for communicating gaps in supply/demand internally and externally
d. Significant challenges
QGI Consulting |5
Railway Demand Forecasting and Service Planning Processes
March 2010
3. Demand Forecasting
Many railway planning processes including financial, asset, capital investment and service planning are predicated on
the railways’ estimate of the freight volumes (demand) to be handled within a given time period. For this information
to be meaningful and support the railways’ planning processes demand forecasts must not only provide an estimate of
the total volumes but also identify the commodities, timing, and the origins and destinations of the traffic. The ability
to view and analyze demand across each of these dimensions allows the railways to plan the type and frequency of
train service, asset requirements including locomotives, freight cars and train crews and to understand the distribution
of traffic across their networks. Each of these views will provide the necessary information to allow the railways to
identify potential capacity constraints and develop strategies to address them. Depending on the nature of the
capacity constraint the railways’ strategies may include capital investment in track and or traffic management systems,
locomotives or rail cars, short term leasing of freight cars and or locomotives, changes to train service or smoothing of
demand peaks through capacity management. The appropriate strategies to be employed will be influenced by the
type of constraint, its duration and where it is expected to occur in the network.
Each year CN and CP create demand forecasts using multiple planning horizons. While the development of the
railways’ annual plans is a principal focus of their planning activities they also develop multi‐year demand outlooks ‐
CN prepares a five year plan and CP prepares a four year plan. Furthermore railway planning activities do not begin
and end with the annual planning process. The objective of the annual demand planning process is to establish the
best view possible of traffic expectations and to define the appropriate service and asset plans necessary to handle
these volumes. Throughout the course of the year both CN and CP continue to do short term tactical planning on a
monthly and quarterly basis to respond to changes in demand by continually revisiting asset and operating plans.
This report describes how CN and CP develop their demand forecasts across the different planning horizons. This will
include: how customer demand information is collected and used, the key “non‐customer” factors considered, and the
process for balancing between customers’ stated demand and the final demand projections included in the railways’
plans. Finally, the report will describe how the information is used by the railways for asset and service planning.
3.1 Long Term Demand Planning
Medium to long term planning activities at CN and CP provide the railways with a highly aggregated view of demand to
allow them to identify potential future changes in key markets that may alter historical traffic patterns and volume
expectations. These demand projections also provide important data for the development of multi‐year forecasts of
financial performance. The long lead times associated with capital investment in the rail industry require them to
develop multi‐year investment plans for asset renewal and expansion with respect to rail cars, locomotives, track
infrastructure and train crews.
6 | QGI Consulting
Railway Demand Forecasting and Service Planning Processes
March 2010
CN
CN’s five year plan is developed each year as part of its annual planning cycle. The plan is based on the detailed
demand forecast and revenue plan for the coming year and is adjusted using a top down approach to account for
medium to long term macroeconomic assumptions and known market issues of significance. Key macroeconomic
assumptions included in the long range outlook will include international, national and regional economic and GDP
outlooks, currency exchange rates and global commodity forecasts for those commodities handled by CN.
The plan is also adjusted to reflect significant market issues that are seen to have a reasonable probability of occurring
within the 5 year planning horizon. This would include items such as new plant start ups, expansions or shutdowns
and mine openings or closures. Finally from a financial planning perspective the railway will factor in its assumptions
for annual pricing yield that is achievable through rate increases.
CP
As part of its annual planning cycle CP develops a four year plan. This planning exercise is focused on eighty (80) key
markets – e.g. Canadian grain exports via Vancouver. Much like CN, CP begins with its base year forecast then
calibrates its plan to factor in macroeconomic assumptions such as GDP projections as well as industry specific
forecasts from independent sources. Macroeconomic data and forecasts are drawn from independent sources
including the major banks and economic and forecasting organizations such as IHS Global Insight.
Key industry inputs include volume forecasts produced by Canada’s major ports for import‐export traffic and industry
specific production and sales forecasts such as for the automobile industry. In addition the railway will factor in
significant market events such as plant closures or expansions.
3.2 Annual Demand Planning
The railways’ annual planning processes include the development of detailed plans for revenues and expenses, freight
car and locomotive fleet sizing, crews, train services and capital investment planning. The foundation for each of these
planning processes is the development of the annual demand or volume forecast for each calendar year.
3.2.1 CN – Annual Planning Process
CN’s annual planning process begins in mid summer with a target of mid fall for completion. While the final demand
forecast results from the combination of bottom up and top down planning it begins with the development of a
detailed bottom up demand plan.
The objective of the bottom up planning process is to develop a detailed and quantified view of the traffic volumes
that are expected to be handled by the railway in the coming year. For planning purposes demand will be expressed
differently depending on the line of business. At CN, with the exception of Intermodal, the principal unit of demand
QGI Consulting |7
Railway Demand Forecasting and Service Planning Processes
March 2010
forecast is carloads. For the railway’s domestic Intermodal business demand is expressed in container units and for
import‐export traffic as twenty foot equivalent container units (TEUs).
There are three principal types of inputs to the demand forecasting process. The initial or base plan is generated using
a combination of the most recent historical and forecast information available. Specifically the starting point is the
current year’s experience which reflects some six months of actual performance and the most current forecast for the
remaining months of the year. The second principal input consists of demand information gathered from the railway’s
customers including any significant issues such as permanent or temporary plant closures, expansions, shifts in
destination markets and seasonal fluctuations in volumes. Finally the railway will apply its own market intelligence
including market share assumptions for individual customers, market forecasts for selected commodities and industry
sectors and macroeconomic forecasts.
Base Year Forecast
The railway’s demand forecasting process should be thought of as an exception based planning process. Given the
breadth and complexity of CN’s market it is not practical to attempt to develop detailed demand forecasts for each of
its more than 2500 shippers, roughly 150 commodities and 13,000 origin–destination pairs.1 At an aggregate level a
significant portion of the railway’s traffic base remains relatively stable from year to year. It is therefore more
practical for these planning activities to focus on how volumes are expected to differ from prior experience rather
than trying to construct a new forecast each year from the ground up. The beginning point is therefore a detailed base
forecast constructed from recent history that is subsequently validated through direct customer discussions and the
analysis of commodity market and macroeconomic forecasts.
CN uses a centralized demand forecasting system that is detailed and hierarchical in structure. At the highest level of
aggregation it provides a view of total system demand by month for the planning period. It also provides integrated
views of the revenue projections associated with projected demand. Underlying this aggregated view are increasingly
detailed views that segment monthly demand across a number of dimensions including line of business, commodity
and commodity subgroup within each business unit, network corridor and ultimately by customer.
1
Canadian originated traffic for calendar year 2007 as provided by CN for the development of a representative sample for transit time
performance measurement.
8 | QGI Consulting
Railway Demand Forecasting and Service Planning Processes
March 2010
Figure 1 Illustrative Schematic of CN Demand Forecasting System
Total System Demand
Forecast information can be input at any level of detail and the planning system will automatically aggregate up or
disaggregate down. Disaggregation of forecast data to the lowest level of detail is done on the basis of historical
traffic movements. Detailed traffic forecasts, including those at the individual customer level, can subsequently be
adjusted manually to introduce more current customer or market information.
Customer Involvement
The principal responsibility for customer interaction at CN, including the collection of demand data, rests with CN’s
account managers that report through its Sales and Marketing organization. CN estimates that customers that
generate approximately 90% of its annual revenues have an account manager assigned to them and that 85% of
demand inputs, excluding regulated grain and Intermodal traffic, can be done at an individual customer level.2
2
If regulated grain and intermodal traffic are included the company estimates that 70% of demand inputs are able to be done at the
individual customer level.
QGI Consulting |9
Railway Demand Forecasting and Service Planning Processes
March 2010
During the planning period CN account managers will meet with their customers to discuss demand projections for the
coming year. This customer interface process is best characterized as informal with the nature of these discussions
varying from customer to customer. Some shippers will provide detailed forecast data on a monthly basis at a
commodity, origin‐destination level that can be directly input to CN’s planning process. In other cases discussions with
customers are less formal with volume outlooks for the coming year discussed in general terms with no specific data
necessarily provided by the customer. The information gathered from customers can then be input to CN’s
centralized forecasting system on an individual customer basis at the appropriate level of detail. In many cases the
information gathered from customers will be used to validate the base plan developed from history with adjustments
made to reflect changes in the timing of shipments, specific commodities to be shipped, and origin‐destination pairs as
appropriate.
Although CN’s planning system provides the capability to enter demand forecast data at the customer‐commodity‐
origin‐destination level for many customers entering data at this level will not be meaningful. This is because
forecasts at this level are now predicting both the behaviour of commodity markets and the market share of individual
customers. It has been the railway’s experience that the level of precision of such forecasts diminishes with each
increasing level of detail and that downstream asset and service planning processes do not require this level of
precision.
There is no formal process for CN to communicate the results of its demand planning process back to individual
customers. Individual account managers may discuss “budgeted” demand with their customers although this process
is informal at best.
Market Reconciliation
While account managers are responsible for gathering most customer information and inputting and validating
customer specific forecasts in CN’s planning system, the development of system wide commodity and commodity
subgroup forecasts is the responsibility of market managers within each of CN’s business units. Individual customer
forecasts are aggregated to the commodity level where they are reconciled by market managers against broader
market views for those commodities. This reconciliation process introduces market share assumptions for individual
customers – in some cases the sum of individual customer market share assumptions can exceed total anticipated
market volumes – given the expected performance of the market as a whole. For instance in developing demand
forecasts for lumber and panel products a market manager would look at market forecasts for Canadian lumber and
panel exports to the United States (US) and projections for US housing starts using independent industry and
economic forecasting data. Once CN establishes the forecast for total Canadian exports of these commodities it
would then weigh individual customer forecasts, historical traffic volumes and competitive factors to arrive at the
expected volumes from each major shipper.
The relative weight given to individual customer forecasts as compared to broader market forecasts will differ across
individual railway business units. Intermodal and grain forecasts for instance will typically place a greater emphasis
10 | QGI Consulting
Railway Demand Forecasting and Service Planning Processes
March 2010
on the expected performance of the total market as opposed to the specific volumes that might be shipped by any one
customer. Projected railway volumes of containerized import traffic that consist principally of consumer products will
be heavily influenced by the expected performance of the Canadian economy as a whole and in some instances
perhaps the expected performance of individual provincial economies. For domestic Intermodal traffic where 85% of
volumes move between CN’s seven largest intermodal terminals but traffic is generated by more than 1,000 individual
shippers understanding forecast demand at a terminal level is more important and practical than trying to predict
shipment volumes and patterns for individual customers when planning asset and service capacity levels. As noted
earlier, for grain shipments it is not necessarily meaningful to forecast traffic volumes at a customer‐origin specific
level because volumes shipped by any individual grain company or processor from any given location will be
determined by regional crop yields and in the case of wheat and barley the export sales programs controlled by the
Canadian Wheat Board.
Top Down Planning
The top down forecasting process done by CN’s Financial Planning group is designed to validate and challenge the
detailed planning assumptions used by Marketing. The top down process uses the preliminary demand forecasts
developed by Marketing and examines at an aggregate level the year over year growth projections as well as
developing trends or significant changes to historical shipment patterns. The plan is then validated against
independent economic and commodity market forecasts to test for reasonableness. The final volume forecast to be
used for corporate budgeting purposes reflects a reconciliation of the bottom up and top down processes arrived at
through discussion between the Financial Planning and Marketing departments.
3.2.2 CP – Annual Planning Process
CP, much like CN, does its annual business planning between August and October each year. CP’s integrated planning
process is known as the Sales and Operations Planning Process. The first step in the annual planning process is the
development of the company’s demand forecast for each calendar year. This forecast is the foundation for all other
components of operational planning through to ongoing plan management. Figure 2 below provides a high level
summary of each step of the planning process.
Figure 2 CP Sales and Operations Planning Process
• Translation of market forecast to workload forecast • Operational workload units
Demand
• Create operational demand views for supply planning • Gross ton miles, train miles, train starts
Planning • Corridor capacity analysis • Potential network constraints
Sales Forecast
CP’s marketplace is similar to CN’s with more than 2500 shippers moving more than 100 commodities between
approximately 14,000 origin–destination pairs.3 For practical reasons CP, like CN, also uses historical volumes and
shipment patterns to create its base forecast prior to incorporating customer and market specific information for the
current planning horizon.
The demand forecasting process looks at CP’s three principal lines of business – Intermodal, bulk and merchandise.
For planning purposes the company will forecast its business using different units of demand, such as carloads or tons,
depending on the specific business unit or market segment. The principal inputs to the Sales Forecasting process
3
Canadian originated traffic for calendar year 2007 as provided by CP for the development of a representative sample for transit time
performance measurement.
12 | QGI Consulting
Railway Demand Forecasting and Service Planning Processes
March 2010
include customer specific demand information and independent market and economic forecasts. Once the
preliminary detailed forecast has been developed demand values are entered into CP’s centralized revenue planning
system (RPS) at the commodity‐corridor level. Corridors define the origin‐destination combinations for the forecast
traffic although they may be defined at varying levels of detail – e.g. station to station, region to region, etc. The data
entered into the RPS system subsequently drives downstream operational and financial planning processes.
Customer Involvement
As noted earlier, with more than 2500 shippers across its Canadian network that ship anywhere from one car to
250,000 cars each, it is not practical for CP to try and engage all customers in demand planning discussions. However,
the company does look to have explicit planning discussions with its largest customers in each of its market segments.
CP has established a threshold of $1 million in annual revenues to identify the customers with whom explicit planning
discussions are to be held annually as part of its sales forecasting process. It is estimated that this represents
approximately 100 customers across all lines of business. These 100 customers would be responsible for well over
80% of CP’s total annual freight volumes. Customer discussions are managed by CP’s Marketing and Sales
department.
There is no formal process for CP to communicate the results of its demand planning process back to individual
customers. Individual marketing representatives may choose to discuss “budgeted” demand with their customers
although this process is informal at best.
Market Reconciliation
The initial view of forecast demand that is created using history and subsequently adjusted for specific customer input
is then validated by CP’s individual market and commodity segment groups. This validation exercise relies on the
knowledge and experience of CP’s marketing personnel and the use of independent industry, market and economic
forecasts. The objective of this process is to determine what the company expects to handle for a given commodity or
market segment and then reflect this in individual customer forecasts. In making assumptions regarding individual
customer volumes CP assesses the probability that an individual customer will achieve the level of traffic they
anticipate given the expected performance of the market as a whole. The validated plan becomes the preliminary
view of demand that is used as the principal input to downstream operating and financial planning processes.
3.3.3 Revenue and Capital Expenditure Planning
Revenue Planning
The demand forecast establishes the basis for the development of the railways’ revenue projections. Revenue
budgets for both railways are essentially developed by applying freight rates per unit of demand (carloads, intermodal
units, tons) against forecast demand. Estimated rates will, depending on the level at which they are applied against
demand, consist of either rates for a specific customer‐commodity‐origin‐destination forecast or weighted average
rates in cases where revenues are estimated against a more aggregated demand forecast. Revenue projections will
also incorporate the companies’ yield strategies and targets for revenue growth established by their respective
executive teams. These yield, or pricing, strategies must take into account the railways’ competitive strategies versus
their direct rail and indirect modal competitors. Railway pricing strategies will also need to reflect the expected
financial performance of their key customers and the effect of freight rates on their customers’ competitiveness in the
customers’ final markets – particularly in markets where market and product competition are important factors.4
Capital Expenditure Planning
Principal areas of railway capital investment include track infrastructure, rail cars, locomotives, facilities and
information systems. These investments are made to achieve a range of corporate objectives including safety of
operations, market growth and operational efficiency or productivity.
The railway marketing organizations’ principal involvement with capital investment planning relates to investment in
freight cars, containers and customer facilities such as commodity trans‐load facilities, to support revenue growth. For
investments such as these the volume of traffic and related risks as well as the associated revenues and profitability
are key inputs to the internal business cases developed when seeking capital expenditure authority. For Operations
departments, capital expenditure planning activities focus primarily on investments in basic plant renewal and
expansion including track infrastructure and traffic control systems, locomotives, railway facilities including yards and
repair shops. These investments will be made to maintain and enhance safety of operations, improve productivity of
operations and thus reduce operating costs or to expand network capacity. All these investments tend to be multi‐
year programs although forecasted demand can influence the timing of these investments based on shifting traffic
patterns that may result in capacity constraints or capacity surplus in specific parts of the network. Where surplus
capacity is forecast in the network, the railway can pursue strategies to salvage track or track materials from sidings or
main track and in the most extreme examples of surplus capacity they can pursue rail line sale or abandonment
strategies.
3.2.4 Challenges
As was noted in QGI’s report on stakeholder operating practices5 supply chain planning requires stakeholder
collaboration with respect to demand forecasting and capacity planning. While shippers have expressed the view
that they believe railway demand forecasts do not always reflect the true demand they communicate to the railways
the railways argue that they must weigh the input of shippers against the expected performance of broader markets
and consider potential volatility in demand when planning capacity and assets in order to manage the financial risks
4
For more discussion of the dynamics of railway competition see: Description of Canada’s Rail Based Freight Logistics System.
QGI Consulting. November 2009. pp 56-63
5
Analysis of Operating Practices: Key Issues from Stakeholder Interviews Potential System-wide Solutions, QGI Consulting October
2009.
14 | QGI Consulting
Railway Demand Forecasting and Service Planning Processes
March 2010
associated with these investments and ensure a commercial return will be achieved over the life of the investments
being considered.
QGI’s interviews with railway representatives revealed that many believe that some of their customers have difficulty
providing accurate medium to longer term demand forecasts. Railway staff identified many reasons for this including:
• inefficiencies in customers’ logistics management processes;
• unrealistic or inaccurate market forecasts and market share projections; and
• an unwillingness of customers to reveal their own marketing projections.
In particular railway representatives indicated that they believed that shippers often had more accurate forecasts
available within their own organizations but that railway representatives simply were not able to access this
information through their regular contacts in the shippers’ organizations. Some railway representatives also believe
that customers may have an incentive in certain situations to inflate their traffic volume forecasts in order to ensure
that railway investment will be made at levels exceeding demand, in order to ensure shipper access to capacity during
times of constraint due to high demand for customers’ products or due to operating disruption that limits available
transportation capacity.
In addition, the timing of annual railway planning activities is, in many cases, not aligned with the timing of their
customers’ internal planning activities. Whereas CN and CP look to begin planning for the coming year early in Q3
many of their customers’ planning activities do not begin until early or mid fall, about the time the railways are
targeting to complete their planning.
At an aggregate level railway forecasts of demand are accurate within approximately 10% over the course of a year.6
An additional challenge for railways is to effectively manage day to day variability in demand that results from the
individual decisions of shippers regarding the timing of their shipments. The railways’ core service design7 does
assume a day of week distribution for demand across its network based on historical movements and uses this to plan
its daily train service. While history is a reasonably good predictor of traffic flows shipper behaviour can change during
the course of year for any number of reasons that the railway cannot anticipate when it establishes it base service
plan.
Shippers in all business segments do not dispute the significant challenges they face in developing reliable forecasts of
transportation demand. However, they point out the special difficulties they face in predicting not only their own, but
also their competitors’ and their customers’ expected behaviour, capacity and requirements in developing such
forecasts.
6
Analysis of Railway Fulfillment of Shipper Demand and Transit Times, QGI Consulting March 2010.
7
Excludes bulk traffic moving in unit trains that while included in annual planning for capacity purposes are not included in the railways’
basic service design as unit trains are proposed and activated based on monthly and weekly planning activities.
Forecasting accuracy can be further complicated when issues of competition and market concentration are
considered. While great effort may be expended by both railways and shippers in the development of detailed
demand forecasts the volume of traffic ultimately handled by the railways will be impacted by modal and intra modal
competition as well as competitive forces in their customers’ markets. Generally speaking the more concentrated a
market segment is, coal or sulphur for instance where there are only a handful of shippers, the easier it is to develop
demand forecasts at the customer origin level. However, in some business segments such as Intermodal, where there
are a very large number of small and infrequent shippers it is difficult and in many cases not practical to attempt to
forecast traffic volumes for individual customers. The accuracy of demand forecasts will also be impacted by broader
market forces – such as occurred during the global financial market upheaval of 2008‐2009. Customer demand for rail
freight transportation fell precipitously during late 2008 however; it would be unreasonable to criticize railway
customers (or railways) for failing to properly forecast the severity of the reduction in freight demand that resulted
from these extraordinary market forces.
Finally, unplanned disruptions in railway operations can lead to changes in the level or timing of rail freight demand.
Examples include weather related disruptions such as occurred during the winter of 2007‐2008 when railway service
was severely disrupted numerous times between December and March. When railway operations are disrupted
some unfilled transportation demand will be shifted to the future creating a higher demand for transportation in
future periods. Determining the degree to which unfilled demand can be shifted to the future and the degree to
which the economic opportunities are lost is difficult to predict and a source of contention between railways and their
customers. However, such disruptions can occur off the railway network as well, particularly when unplanned
production shutdowns due to strikes, plant operations problems, or ocean transportation disruptions result in changes
to freight logistics demand levels and timing.
16 | QGI Consulting
Railway Demand Forecasting and Service Planning Processes
March 2010
4. Service and Asset Planning
Identifying and forecasting the volume of traffic that is expected to be handled during a year is the initial step in the
railways’ planning processes. An important output of demand forecasting activities is a series of different views of
forecast traffic volumes that are used in railway asset and service planning processes. The following section of the
report describes how market demand forecast information is transformed and used for railway operational planning.
4.1 CN Annual Service and Asset Planning
As described earlier, CN’s principal unit of demand is the number of rail cars the railway plans to move during the
course of a year. However, this unit of demand alone is not sufficient to meet the needs of CN’s downstream asset
and service planning activities. As such the carload forecast is translated into other units of demand including tons
and revenue ton miles (RTM). 8 Tonnage forecasts are derived from the carload forecast by applying historical ton per
car ratios at the individual commodity‐car type level.9 These forecasts are then used to calculate revenue ton miles.
RTMs are a measure of revenue workload calculated by applying the average length of haul against forecast tons for
all movements at a commodity‐origin‐destination level. At the most granular level this calculation is done for an
individual shipper‐commodity‐origin‐destination combination. Where traffic is forecast at a higher level of
aggregation such as commodity‐origin‐destination CN’s planning system will calculate RTMs using the average
weighted historical length of haul for all of the forecast traffic included at a particular level of aggregation.
4.1.1 Rail Car Fleet Planning
Forecast carloads are used to plan CN’s rail car fleets. This unit of demand is seen to be effective and appropriate for
this planning activity because it is accurate at an aggregate level. It is not however sufficiently accurate at a detail
level such as the origin‐destination flow level to be effective for train service design planning as this requires more
accurate estimates of the tonnages to be moved.
In developing its rail car fleet plan CN’s objective is to define the total number of cars by individual car type that it
believes will be required to handle the forecast traffic. The number of cars required to handle a defined volume of
traffic is dependent on a number of variables including: the absolute volume of traffic to be moved and the planned
car cycles10 for each individual car fleet. In situations where rail cars are used in single point to point dedicated
8
Revenue ton miles are also referred to as net ton miles.
9
CN’s planning system associates a specific car type against all forecast commodities.
10
A freight car cycle is a measure of the elapsed time for the movement of a car including the time for each segment of a car trip. The
segments include time in loading, loaded transit, unloading and empty transit to its point of next loading. Car cycles for individual fleets
are generally computed by summing the time spent by all cars for all car trips in each cycle segment and dividing by the total number of
car movements in that segment. The weighted average times for all segments are then added together to produce an average car
cycle number.
service the sizing of a rail car fleet is relatively straightforward. However, with the exception of specialized equipment
operating in customer assigned pools this is rarely the case. Rather the railway’s car fleets are largely used in general
freight service and are shared by many shippers for movements between many origins and destinations with varying
lengths of haul with highly variable cycle times. As such the railway uses average cycle times for a given fleet based on
historical traffic distribution patterns.
The principal steps in sizing each rail car fleet can be summarized as follows:
1. Develop demand forecast expressed in carloads
2. Identify specific commodity movements within the forecast for which the railway will supply the cars
3. Calculate average car cycles by car type based on historical traffic movements
4. Adjust historical car cycles to reflect asset utilization targets (car cycle improvements) by car type
5. Apply car cycle data by car type against forecast demand to determine the number of rail cars required to
handle the traffic through the course of the year accounting for seasonal fluctuations in traffic
6. Compare calculated car requirements against existing fleet positions including planned car retirements,
expiring leases during the planning period, target bad order ratios11, planned rail car maintenance or
modification programs and assumptions regarding substitution12 of cars between fleets
7. Identify forecast fleet shortages and or surpluses by car fleet
In cases where the fleet planning process yields a projected car shortage for a given car fleet CN will assess the
strategies available for eliminating the shortage considering the size and duration of the projected shortage. Options
generally include the lease or purchase of additional cars. Because of the long lead times associated with new car
construction, typically one year or more, this option will not be considered unless the shortage is projected to be long
term and a business case can be developed to support the investment. Leasing options will be limited by the
availability of cars in the North American freight car market and the applicable lease costs. Given the homogeneous
nature of most rail car fleets in North America if a car shortage is being driven by a general increase in market
volumes, in grain for instance, the lease market may also present limited options as other railways may also be
pursuing the same car types at the same time.
Where the rail car planning process identifies a projected surplus position the railway will seek to minimize the costs
associated with the surplus asset through leasing or sales opportunities with other railways or rail car leasing
companies, storage of cars or returning leased cars to lessors if it is financially advantageous.
11
The term bad order is used to describe a rail car or locomotive that is removed from service due to mechanical failure. In fleet
planning for both rail cars and locomotives railways assume a certain level of ongoing bad orders or the average percent of a fleet that is
unavailable for service on a daily basis. (Bad order ratio = cars unavailable / total cars in fleet) The bad order ratio applied for a given
fleet of cars or locomotives will typically reflect the average age of the fleet, its mechanical history and targets for asset utilization
established by the company.
12
While some freight movements require very specific car types to handle the traffic, other traffic can be moved with a variety of car
types. For example, while some woodpulp shippers may request cars of a specific weight capacity with a specific door type, they may
accept substitute cars with lower weight capacity and different door configurations. Similarly, users of covered hopper cars may prefer
cars with specific types of top hatches and unloading gates however railways may substitute cars that do not have the preferred loading,
unloading and capacity configurations.
18 | QGI Consulting
Railway Demand Forecasting and Service Planning Processes
March 2010
4.1.2 Service Design Planning
The objective of the service design or train service planning process is to develop the detailed base train service plan
required to handle the forecast traffic volume and to accommodate where required scheduled passenger train
operations. The output of this planning process is a system wide base train service design that identifies the schedules
and frequency of trains that the railway plans to operate during the planning period. At a detail level the train service
plan will identify the operating parameters for each scheduled train service including:
• the days of the week it will operate and scheduled departure and arrival times;
• the origin and destination terminal for each train;
• the type of traffic planned to be moved including how traffic will be blocked (grouped on the train) to facilitate
the setting off and picking up of traffic at intermediate terminals for delivery to customers and to make planned
train connections;
• established priorities for traffic in cases where available traffic exceeds available train capacity;
• on line work (switching) to be performed by each train; and
• maximum train lengths and weights based on routing to account for grades, curvatures, maximum gross weight
on rail restrictions and siding lengths to allow for train operations in both directions on single track territories.
In addition to scheduled train operations, the railway needs to plan for the movement of trains that operate on
demand, in response primarily to bulk shippers’ requirements. Most unit train movements operate in response to the
production and marketing requirements of bulk shippers. The demand for this type of traffic is usually defined in
terms of the number, tonnage and weight and length of unit train movements that will need to operate in specific
corridors on a monthly and weekly basis. The demand for these train movements may not be scheduled in advance
for particular days and thus may need to take advantage of capacity “windows” on the railways’ networks in the
corridors where they will operate. Thus the service design plan will need to provide for these capacity windows and
anticipate how the expected volumes of unit train traffic will be accommodated taking into account the needs of
regularly scheduled train service.
Finally, the railways’ service design teams will need to forecast and plan for time periods when tracks may be removed
from service or have restricted operations to allow for scheduled maintenance operations to take place on track, track
materials, right of way or traffic control systems.
The principal demand input to the service design planning process is forecast gross ton miles (GTM). Gross ton miles
are a measure of railway workload calculated as the number of gross tons including freight cars and their contents,
company service equipment, and cabooses multiplied by the distance moved – e.g. 10,000 gross tons x 1 mile = 10,000
gross ton miles. For planning purposes GTMs are calculated based on the revenue ton mile forecast using historical
GTM/RTM ratios of approximately 2:1. For train service design GTMs rather than RTMs are used as GTM calculations
will include the weight of the rail cars and empty car movements that must be accounted for when calculating train
capacity, locomotive requirements and track maintenance. Figure 3 below provides a schematic view of how the GTM
forecast is created.
QGI Consulting |19
Railway Demand Forecasting and Service Planning Processes
March 2010
Figure 3 GTM Forecast Development Process
Monthly Forecast Carloads The forecast GTMs are then applied against the
(by commodity, origin‐destination)
base service design at a detailed level using
Apply historical ton per car
ratios by individual car type
historical traffic routing patterns.
Monthly Gross Ton Miles (GTM)
(by commodity, origin‐destination)
4.1.3 Locomotive Planning
With its core service design in place CN then plans the locomotive requirements associated with the anticipated train
operations. Locomotive planning at CN is done in three incremental phases that define the locomotive requirements
for each line of business and service type. The three planning phases are:
• the base service plan including all manifest and intermodal train service
• bulk train service including coal and sulphur moving in unit trains
• grain train service including unit train and country spotting operations
CN uses train service simulation models to identify the demand for locomotives associated with operating its core
service design. These simulations identify the locomotive requirements for each train run and allow CN planning
personnel to identify the time locomotives spend hauling trains versus the time spent in yards. By looking at inbound
to outbound train connections CN can identify the most efficient way to cycle locomotives between trains in order to
minimize terminal dwell time and obtain the most efficient utilization from its fleet. As with rail car planning CN will
factor into its locomotive demand projections a planned bad order ratio. This yields the planned locomotive supply for
the base service plan.
Locomotive demand for forecast coal and sulphur traffic unit train service is based directly on the commodity demand
forecast. The forecast demand is converted into planned train runs based on an expected number of cars per train.
Each train has a planned train cycle time that will include loading at origin, loaded transit, unloading at destination and
20 | QGI Consulting
Railway Demand Forecasting and Service Planning Processes
March 2010
empty transit back to origin. CN plans for this traffic to move in unit train service with dedicated locomotives for each
train operated. The number of locomotives required therefore will be based on the number of locomotives assigned
to each train multiplied by the number of trains divided by the planned train cycle.
For grain traffic locomotive demand planning takes into consideration grain service type – grain traffic moving on
manifest trains, in unit train service and for country spotting and pick up operations. Grain traffic moving in manifest
service will have been included in the base service plan and is therefore excluded here.
4.1.4 Train Crew Planning
In planning its train crews CN needs to account for two different wage structures – hourly and mileage based crews.
Train crews in Quebec, on the former BC Rail territory, and in the United States operate under hourly wage
agreements with all other train crews in Canada under mileage agreements. The total number of train crews
required is determined by calculating the number of train miles (mileage based crews) and train hours (hourly based
crews) required to execute the operating plan including scheduled train services and unit train service. Once the base
requirement is established the plan is adjusted to factor in expected sick days, vacation time and planned retirements.
4.2 CP Annual Service and Asset Planning
As discussed earlier CP employs an integrated annual planning process called the Sales and Operations Planning
Process. This process encompasses all planning activities from the development of the initial demand (sales) forecast
through to the execution of the integrated operating plan. The demand, supply and production planning modules of
this integrated planning structure are where asset and service planning activities happen. In interviews with QGI, CP
indicated that the purpose of its operational planning process is to create a balanced network plan while maximizing
train length in order to ensure effective asset utilization and optimize the capacity of its network.
4.2.1 Operational Demand Planning
The operational demand planning phase uses the sales forecast as its key input and translates market demand
expressed in tons into units of workload demand that can be used for train service and asset planning. There are three
keys methods employed in this phase of planning – forecasting of train miles, train start modeling, and product design
and analysis. The planning process uses all three methods to estimate the operating workload and rationalizes and
brings them together into the Integrated Operating Plan.
Train Mile Forecast
CP, like CN, translates its market demand forecast expressed in tons into revenue ton miles by applying the average
length of haul for movements at the commodity‐corridor level. GTMs are subsequently calculated from the RTM
forecast using historical ratios. Lastly train miles are calculated from the GTM forecast by applying historical
GTM/train mile ratios. Underlying these calculations is the core train design. The key output from this process is the
total estimated workload for CP’s network stated in train miles by service area and type of train service. Because this
view of workload assumes the historical train service it is subject to adjustment if changes are made to past train
design.
Train Start Modeling
In parallel with the train mile forecast process is the development of planned train starts. Train start modeling is done
for approximately 80 corridors and 15 major terminals in CP’s network. Train starts are calculated from forecast
carloads from the sales forecasting process. Total carloads to be handled, both loaded and empty, are calculated by
applying historical empty‐load ratios to the forecast for loaded cars. The number of trains required is then calculated
by applying historical average car lengths and train length/weight ratios against the total number of cars. The
principal output of this process is an estimate of the total number of trains required to meet the demand in the
forecast expressed in terms of total monthly train demand by network corridor. Estimated train starts at an individual
corridor level are then compared to corridor capacities to identify potential constraints.
Product Design and Analysis
Train service design planning is done using CP’s internally developed Product Design and Analysis Visibility Tool (PDAE)
in conjunction with the MultiRail13 train service planning application. The objective of this planning exercise is to
assess the capability of the existing service design to adequately handle the forecast traffic volumes. The current
demand forecast is translated into a detailed traffic file by applying the forecast volumes against a historical traffic file
thus providing a view of forecast demand by origin‐destination station by day of week. This traffic file becomes the
input to the MultiRail system that CP uses for service design planning. The detailed “forecast” traffic file is loaded to
the MultiRail system which enables CP planning personnel to assess the ability of the existing train service design to
accommodate the forecast demand. The system identifies opportunities for changes to the existing design to optimize
handling of the forecast volumes.
4.2.2 Supply Planning
The supply planning phase looks to size the assets required to execute the service plan. This includes rail car and
locomotive fleet planning and train crew planning. All asset planning builds in defined productivity initiatives.
Locomotive fleet sizing uses the MultiRail Locomotive planning module. This module identifies the locomotive demand
associated with the operation of the service design.
Rail car fleet planning is done by CP’s car management team. Car management uses the monthly demand forecast
from the sales and revenue forecast to plan fleet requirements for seven major car fleets consisting of twenty sub
fleets. Car requirements are calculated by applying average monthly car cycles adjusted for productivity initiatives and
13
MultiRail is a third party train service planning software developed by MultiModal Applied Systems of Princeton, New Jersey.
22 | QGI Consulting
Railway Demand Forecasting and Service Planning Processes
March 2010
for seasonality against the forecast demand. CP also factors in an average 4‐6% bad order factor based on the
mechanical history of individual car fleets. Some fleets will have higher maintenance requirements and thus
corresponding bad order ratios based on the types of commodities they handle and the loading and unloading
practices of shippers and receivers.
Crew planning is done by CP’s crew management team. Crew demand projections are based on projected train
demand by corridor. The number of crews required is calculated by determining the crewing requirements for each
forecast train run within each individual corridor. Train crew requirements are adjusted for planned absentee rates,
sick leave and vacation time.
4.3 Annual Service and Asset Planning Challenges
The biggest challenge to service planning for CN and CP is the volatility of demand on a day to day basis. While
planning activities are detailed and use the best available information regarding expected traffic volumes they cannot
realistically account for the inevitable ebb and flow of daily traffic volumes resulting from the decisions of individual
shippers, the actions of other supply chain participants or unanticipated events in either the railway network or
customers’ markets. Railways recognize that perfection in this environment is unachievable as there are too many
factors that are not within the railways’ control. The objective of these planning processes is to establish a core
service plan, supported by sufficient assets to meet expected demand with sufficient flexibility to respond to day to
day occurrences at an acceptable level of financial risk to the railways.
4.4 Short Term Planning and Asset Management
The annual planning processes at CN and CP establish the volume and revenue expectations for the coming year and
contribute to the development of operating plans that drive expense forecasts and projections for corporate financial
performance. Once finalized these individual plans are translated into financial budgets and workload projections for
the year and become the basis against which actual performance is measured. While the annual planning process
comes to a close at this point both CN and CP continue to do short term planning on a monthly and quarterly basis to
measure performance to budget and adjust the outlook for the remainder of the year regularly revisiting asset and
operational plans.
Although significant time and effort is expended in creating the annual plan a forecast created 3‐4 months prior to the
beginning of the year cannot perfectly anticipate evolving market conditions and customer behaviour particularly in
times of market volatility. As such both railways revise projections for revenues and volumes using monthly and
quarterly outlooks. These outlooks seek to incorporate the most current customer and market information to confirm
expectations for the balance of the year and then translate revisions into refreshed views of asset and train service
requirements going forward.
The question the railways are continually attempting to answer is whether or not performance to date, whether above
or below plan, is a leading indicator of volume and financial performance to follow. In cases where demand has fallen
short of expectations should the railway expect to see a “snowplow” effect whereby deferred demand will materialize
later in the year? Other possibilities are that the demand expired as it was either time sensitive or was satisfied by
other suppliers or via another transportation provider and is effectively lost to either the shipper, the railway or both?
Similarly, in situations where demand has exceeded planned volumes to date ‐ is this an indication of a sustainable
increase in demand that will carry through to the end of the year; has demand projected for later in the year simply
moved forward; or will volumes return to planned levels? These are important issues for the railways to manage as
they can have a significant impact on rail car, locomotive and train service plans that have been put in place using the
original plan assumptions.
Monthly and quarterly planning processes, including discussions with their customers, attempt to provide the best
possible short term outlooks that can be used to make decisions regarding changes to rail car and locomotive fleets
and whether or not the existing train service plan is suitable going forward. In addition to trying to maintain service
levels to customers and ensure sufficient capacity is in place the railways are looking to maintain control of their short
term operating costs.
Both CN and CP have implemented formalized processes for reviewing current year performance and translating
changes in demand outlooks into operational workload projections that allow for the ongoing review of asset and
service plans.
4.4.1 CN Short Term Planning Processes
CN’s Equipment Requirements Team (ERT) is responsible for conducting ongoing reviews of the company’s asset
requirements and plans throughout the year. The team is comprised of executives from the Marketing, Financial
Planning and Operations Planning groups and is chaired by the vice president of Financial Planning. The ERT meets
monthly to review asset plans for both marketing and operations based on performance to date and revised outlooks
going forward.
For rail car fleet planning the monthly ERT process is supported by a weekly process that tracks forecast changes in
short term customer demand as identified by Marketing and actual realized demand as measured through the
company’s car ordering processes. With a view of projected demand and current actual demand on rail car fleets as
well as fleet performance (car cycles) the process projects potential surpluses or shortfalls for rail cars by individual car
type. Also included in this analysis are measures of recent and short term fleet attrition including retirements, cars
destroyed, and expiring leases. Using all of these factors the ERT process determines actions to be taken to meet
expected demand within expected performance levels in consideration of established expense targets. Decisions of
the ERT are required to be unanimous and require sign off at the VP level in each function.
24 | QGI Consulting
Railway Demand Forecasting and Service Planning Processes
March 2010
Locomotive Planning
CN continually assesses the performance of its locomotive fleet to identify situations where short term locomotive
capacity is becoming constrained or surplus. CN uses its internally developed Locomotive Planning System (LPS) to
measure both the daily performance of its locomotive fleet and the near term demands on planned train operations.
This is done by loading the current core service design and planned bulk train demand into the LPS which provides
CN’s management team with a dynamic view of forecast daily supply and demand for locomotives at the individual
terminal and train level.
As part of the locomotive management and planning process CN monitors the performance of a number of key drivers
including: rail car dwell times, train length, and car velocity (average car miles per day per car). The performance of
each of these operational elements is used to identify emerging trends that may signal the need for additional
locomotives or alternatively a surplus fleet. For terminal dwell times the focus is on identifying cars that have dwell
times greater than 24 hours. An increase in the number of such cars may signal the need to either operate additional
trains or longer trains in order to avoid a build up of cars in terminals. Operating additional or longer trains can
increase demand for locomotives. Similarly average car velocity is used as an indicator of network efficiency and
fluidity. If performance falls consistently below established thresholds it may signal the need to activate additional
trains. The railway’s train service design assumes a planned train length and weight and a locomotive horsepower to
train weight ratio. Short term fluctuations in demand can result in less or more traffic being available for a designed
train service that may result in longer or shorter train lengths than originally planned. By tracking this performance CN
is able to identify developing trends and make tactical adjustments to its locomotive distribution strategies. When
train length is decreasing it may present the railway with opportunities to combine trains thus operating fewer total
trains and reducing the demand for locomotives. When train length is increasing it may signal the need to run
additional trains thus placing greater demand on the locomotive fleet.
Where short term locomotive shortages or surpluses are identified they are typically addressed using locomotives
made available from or to other railways or locomotive leasing companies through either short term leasing or trading
of horsepower hours. This latter approach is relatively common in the North American railway industry. Individual
railways establish bilateral agreements to use one another’s locomotives for short periods of time. In lieu of
traditional commercial lease arrangements the railways track the use of their respective locomotives using a
debit/credit system. Horsepower hours are assigned a dollar value and agreements may provide for reconciliation of
imbalances between railways at prescribed times.
4.4.2 CP Short Term Planning Processes
CP’s ongoing planning activities and performance oversight are managed through its internal “Butterfly Team”
structure. This group, which exists in parallel at the senior management and executive levels within the company, has
direct responsibility for measuring performance to plan and conducting ongoing assessments of railway capacity issues
throughout the course of the year. At the working level the Butterfly Team consists of representatives from all key
Marketing and Operations functions within the company including product design, fleet planning, locomotive
planning, crew planning, engineering, yield management and revenue planning. The Butterfly Team approach was
adopted by CP Rail in recent years in direct response to capacity issues encountered in its critical western corridor.
The team conducts a monthly review of updated demand forecasts that incorporate the most recent market and
customer intelligence and the company’s performance and experience to date. Using analytical and planning
techniques similar to those used in the development of the annual operating plans the team looks to identify potential
near term capacity constraints in all aspects of network operations by asset class and location. Where potential
capacity issues are identified the team assesses the range of alternatives available and makes recommendations to
CP’s executive team.
A key management tool used by the team is a network status model that describes the capacity condition of the
network based on defined thresholds. Network condition is classified at a high level using color coded descriptions
(blue, green, yellow, red) that describe the state of network capacity from surplus to constrained. The designation of a
network corridor or asset class (crews, locomotives, freight cars) as yellow or red indicates the potential for near term
capacity constraints and leads to the assessment of tactical options available to alleviate the projected capacity issue
based on the specific nature and location of the constraint. Capacity constraints in a railway network can be the direct
result of operational or market issues.
From an operating perspective short term capacity constraints can result from network disruptions including weather
based disruptions and operational disruptions such as train derailments or problems experienced by one of the
railway’s customers at a major bulk or container terminal. The severity of the constraint, and therefore the solutions
to be considered, will be determined by the location and duration of the disruption. A disruption on a low density
branch line may impact local service to some customers but is unlikely to have a broad impact on the network as a
whole. By comparison a main line disruption in a high density corridor such as CP’s western corridor can have
significant impact not only in the immediate area but if of sufficient duration will have ripple effects in other parts of
the network. In addition to disrupting corridor train operations this kind of disruption can impact the flow of empty
cars to meet future orders, reduce rail car fleet capacity by lengthening car cycles, impact crew and locomotive
balancing and create congestion in railway terminals throughout the system as the predicted flow of traffic is
disrupted.
Short term market shifts can also create capacity issues in the railway network. While most often thought of in terms
of capacity constraints significant shifts in market volumes can also result in a short term capacity surplus. In one case
26 | QGI Consulting
Railway Demand Forecasting and Service Planning Processes
March 2010
the railway will look at solutions to try and increase short term capacity and in the other ways to reduce capacity in
order to control operating costs and maintain efficient operations. Bulk commodity markets such as coal and grain are
good examples of where changing market demand can result in both of these situations.
Some examples of tactical responses that would be considered by CP’s Butterfly Team in these situations are
illustrated in the following table.
Freight Cars • Add to existing fleets by activating cars in • Reduce car fleets through short term storage
storage or through short term leasing of surplus cars
• Allocate or ration cars to customers
Crews • Increase train crew pools through recall of • Reduce train crew pools through short term
laid off employees if capable lay offs
5. Summary
The processes for demand and service planning at CN and CP, while somewhat different at a technical level are very
similar with respect to the specific areas of planning, the key inputs used and the planning horizons employed.
Each year both railways plan their expected volumes of traffic and the assets and train services needed to support
them using a number of different planning horizons. The focus of most planning activity is centered on the
development of the railways’ annual plans. These plans form the basis for longer term planning activities – for CN a
five year plan and for CP a four year plan – that look to identify significant expected changes in railway volumes, traffic
patterns and financial performance by introducing medium term macroeconomic assumptions and industry specific
forecasts using a top down approach. These longer term views of demand are important to the railways as they are
critical inputs to each company’s multi‐year capital investment strategies for asset renewal and capacity management.
The annual demand planning process is the initial planning activity and produces the forecast traffic volumes that
become an important input to the downstream financial and operational planning activities for both railways. These
demand forecasts are created using a combination of historical traffic performance, direct customer input and market
intelligence. The railways differ somewhat with respect to the how they engage their customers and how many
customers they look to involve in planning discussions. However, for both railways there is tremendous concentration
of traffic volume within a small percentage of their shippers. In general, on both railways, over 80% of rail volumes
are moved by approximately 5% of shippers. Thus, railways can reliably plan their asset requirements at a high level
using the input of a relatively few shippers.
Key operational planning activities for both carriers are focused on sizing of rail car and locomotive fleets and the
development of train service plans at levels necessary to meet demand projections including the anticipated timing of
demand. In planning these assets both carriers explicitly factor in assumptions regarding the mechanical reliability of
their fleets and productivity initiatives.
The railways share similar challenges in developing accurate demand forecasts related to customer and market factors
beyond their control. With respect to customers the railways sometimes find it difficult to engage customers in
planning discussions because their respective planning timelines are not necessarily aligned. In some instances the
accuracy of demand forecasts provided by customers is suspect because of the individual market share assumptions
used – a key reason why the railways explicitly validate individual customer forecasts against broader market
forecasts.
Unforeseeable short term volatility in global commodity markets also present challenges for railway demand
forecasting as they do for the railways’ customers. In addition the volatility of day to day demand driven by the
decisions of individual shippers and the actions of other supply chain participants can be problematic for the railways.
The railways necessarily design their core train service using assumptions, based on history, of how volumes will flow
28 | QGI Consulting
Railway Demand Forecasting and Service Planning Processes
March 2010
on a daily and week to week basis. While history is a reasonably good predictor of traffic flows, shipper and other
stakeholder behaviour can change during the course of year for any number of reasons that the railway cannot
anticipate when it establishes it base service plan.
CN and CP recognize, as do many shippers, that annual demand and service planning is not perfect and should not be
expected to be. The challenge facing the railways is to establish operating plans that provide sufficient flexibility to
reasonably adjust to short term market fluctuations and unanticipated disruptions within the logistics network at an
acceptable level of financial risk. It is for this reason that both CN and CP expend significant effort in monthly and
quarterly planning activities. The railways attempt to compensate for these types of events by continually measuring
their performance to plan and assessing and adjusting their operations to maintain service consistency for customers
and fluidity within their own networks.