Asset Velocity

Asset Velocity, Terminal Performance and Network FluidityThe critical role of terminals, protecting the manifest network,and the need to manage and measure more than train velocity

May 24, 2006

C O N F I D E N T I A L

Prepared for: MultiRail User Conference

© 2006 Mercer Management Consulting www.mercermc.com 1

Contents

Importance of Asset Velocity, Cycle Times and Terminal Performance

Railcar Velocity and Cycle Time

Terminal Performance

Locomotive Velocity and Cycle Time

Manifest Traffic and Network Fluidity


Importance of Asset Velocity, Cycle Times andTerminal Performance

Cycle times are critical in determining:– Total fleet requirements for railcars and locomotives– Customer service levels

As the determinant of fleet requirements, cycle times impact millions of dollars in investment decisions

All equipment cycles are made up of a series of processes:– For railcars these include:

- Shipper load time- Terminal and yard time (loaded & empty)- Train movement time (loaded & empty)- Consignee unload time- Storage time (loaded & empty)- Shop/repair time

Asset velocity can be viewed as an alternate measure of cycle time, but only when it is measured as total distance traveled/total cycle time

– Measuring only train speed leaves out many important components of cycle time that are not dependent on train performance

– Some of the most significant additional components being terminal processing time, terminal wait time, and servicing time

– These additional time components often exceed the time spent in trains

– For locomotives these include:- Servicing & fueling time- Idle time waiting for assignment- Time in held trains- Time moving in trains- Deadhead movement time- Shop time


Consignee12%

Consignor12%

Empty Yard & Term

35%

Empty Movement

7%

Loaded Movement

8%

Loaded Yard & Term.26%

How Cars Spend Their Time:General service rail cars spend the vast majority of their time in activities other than moving on trains.

The relative size of individual cycle time components has varied little over the last 30 years.

Unfortunately, the data to support analysis of cycle time components is no longer publicly available –however there is substantial evidence that the basic factors have remained unchanged for 30+ years.

Source: Reebie Associates, “Toward an Effective Demurrage System,” 1972

Cycle Time ComponentsAll car types in general service

0

5

10

15

20

25

30

1990Boxcars

1980Boxcars

1972 AllCars

Days

Shipper Time Loaded Time Consignee Time Empty Trip Time

Carload Cycle Time 1972 - 1990

Source: Little, Kwon & Martland, “An Assessment of Trip Times and Reliability of Boxcar Traffic,” 1992


Evaluating Equipment CyclesFor manifest traffic, terminal performance is of equal or greater importance compared to train speed

For any equipment cycle, part of the cycle time is spent in trains, and part is spent in other activities

Train speed determines only one or two of the time components that make up the overall equipment cycle, and thus cannot be the sole measure of a railroad’s operational effectiveness

While on-time performance of the trains may influence the time spent in non-train activities, in general train speed has little impact on the duration of these non-train activities

The relative importance of managing and measuring these non-train activities depends on the percentage of total time spent in each of these activities

For example, if 60% of a general manifest movement is spent in yards and terminals, then managing this time is very important

– By the same token, if only 20% of the time is spent moving in a train, then this time is comparatively less important

Because time spent in terminals represent over half of the cycle time for manifest traffic, terminal performance becomes one of the single most important determinants of cycle time

– Studies undertaken at M.I.T. starting in the 1970’s clearly demonstrated the critical role of terminals in overall cycle time

– They also showed that most trip plan failures occur at the terminals, and thus they strongly influence overall transit time reliability

Similar issues relate to locomotives in terms of servicing time, wait time at terminals, and deadhead time.

Equipment velocity, cycle time, and productivity measures that encompass all of the elements of the equipment cycle are the only true way to determine operational performance (and customer service)


Contents







A Typical Carload Movement

40% of time spent in a train60% spent in a yardResult is an overall 6.8 mph for the car (163 miles/day)5 mph or 8 days transit would be more typicalAdditional 2 days is typically all yard time!

ReleaseFrom

Shipper

LocalServingYard A

SystemYard B

SystemYard C

LocalServingYard D

DeliveryTo

Customer

An aggressive 5 day-20 hour schedule with no failuresActivity Time Day Train Hrs Yard Hrs Distance SpeedShipper Release 1600 Day 1Local Pick-up 1 Day 2 8Arrival at Yard A 600 Day 2 6 75 12.5Local Train from A 1800 Day 2 12Arrival at Yard B 600 Day 3 12 200 16.7Road Train from B 1000 Day 4 28Arrival at Yard C 600 Day 5 20 400 20.0Local Train from C 1000 Day 6 28Arrival at Yard D 2200 Day 6 12 200 16.7Switcher from D 600 Day 7 8Customer Delivery 1200 Day 7 6 75 12.5Totals 56 84 950 6.8


Asset Velocity and Trip Time Levers

For carload, at best, train time represents perhaps 40% of total cycle time, and 15% to 20% is closer to reality

– This means that changing train speed will have little impact on total transit time or railcar asset requirements

– Improving local service delivery and reducing (or eliminating) yard time will have a much bigger impact as it represents a larger proportion of total transit time

– Improving yard connection reliability is one of the best ways to reduce transit time and increase asset velocity

From a customer’s perspective, it is car or shipment velocity, not train speed, that matters

For pure shuttle train operations, train speeds are a more significant component of overall cycle time

– However, operating “single commodity” trains may provide false economies (see later slide)


Trip Time Levers

Consider a 1 mph increase in train speed for all trains– Current composite average speed is 17 mph– A 1 mph increase would reduce transit time by 3 hours

Consider a connection failure at any yard with once-per-day train service– This would add 24 hours to transit time (with no change in train speed)

Typical hump yard connection failure rate is about 25%– Consider 100 cars making a 22 hour connection, with a 25% failure rate– 75 cars will take 22 hours, and 25 cars will take 46 hours, for a 28 hour average– If the failure rate was reduced to 10%, the average would drop to 24.4 hours, saving 3.6 hours

with no change in train speed

Average yard time as reported to the AAR by several major railroads is about 28 hours– A number of yards on these railroads are above 36 hours– Consider a yard with: (a) evenly spread, random car arrivals, (b) once per day, evenly spread

train departures, and (c) a minimum processing time of 10 hours– Such a yard would achieve 22 hours average yard time (10 hours processing, 12 hours on

average spent waiting for a train departure)– Using this simple model, a yard with a 12 hour minimum processing time would need a 50%

connection failure rate to reach 36 hours of average yard time, or a series of cascading failures for a smaller subset of cars

Activity Time Day Train Hrs Yard Hrs Distance SpeedShipper Release 1600 Day 1Local Pick-up 1 Day 2 8Arrival at Yard A 600 Day 2 6 75 12.5Local Train from A 1800 Day 2 12Arrival at Yard B 600 Day 3 12 200 16.7Road Train from B 1000 Day 4 28Arrival at Yard C 600 Day 5 20 400 20.0Local Train from C 1000 Day 6 28Arrival at Yard D 2200 Day 6 12 200 16.7Switcher from D 600 Day 7 8Customer Delivery 1200 Day 7 6 75 12.5Totals 56 84 950 6.8


Single Commodity TrainsOperating “single commodity” trains may provide false economies

In many cases single commodity trains are created by accumulating cars until a full train can be run– A common example is grain trains, but this may apply to other commodities including soda ash,

ethanol, steel products, etc.

Example:– A series of shippers release 30 cars per day, these cars are accumulated over a 3 day period to

create a single commodity train– A set of local trains pick-up the cars, and bring them to a central collection point– The fastest scenario (for the third day’s traffic) is as follows:

One must add 48 and 24 hours of yard time to the above for the 1st and 2nd day’s traffic

This yields an average yard time of 54 hours, and an average transit time of 85 hours– In this example the cars are spending 64% in yards, not too different from our carload example– Also indicating that increasing train speed will not significantly impact asset velocity (or yard space

requirements to hold cars being accumulated)

Activity Train Hrs Yard Hrs Distance SpeedShipper ReleaseLocal Pick-up 8Arrival at Assembly Point 6 75 12.5Train from Assembly Point 22Customer Delivery 25 400 16.0Totals 31 30 475 7.8

AssemblyPoint

DeliveryTo

Customer

ReleaseFrom

Shipper


Contents







Norms for Hump Yard Performance

A large hump yard approximates a situation where one can view arrivals and departures as a statistically random process, evenly distributed across the day

Assuming once per day departures for all blocks, this means that the average yard time for the case where all connections are made, and all trains are operated as expected, should be the yard’s processing time, plus 12 hours

Cutoff or processing times for a hump yard should generally fall in the 8 to 12 hour range, producing a statistical average yard time of 20 to 24 hours

– Based on above we will use 22 hours as the statistical “ideal”average yard time

Less than daily train service (e.g. 5 day/week service), cancellations, and connection failures will push the average yard time up

– A 25% connection failure rate on a 22 hour base, pushes the idealized average yard time to 28 hours (failure causes could include cancelled trains and the need to leave tonnage behind)

Timing high volume inbound/outbound connections and providing multiple departures per day will drive the average yard time down

This generally means that a well run hump yard should have an average yard time in the 18 to 28 hour range (assuming we do notcount stored and bad order cars)

TrainArrival

ProcessingTime

WaitingTime

DelayTime

ForMissed

Connection

TrainDept.

TrainDept.

Time


Impact of Yard Reliability and “Daily Doubles”

Improving yard connection reliability is one of the best ways to reduce transit time and increase asset velocity – as noted earlier:

– Consider 100 cars making a 22 hour connection, with a 25% failure rate– 75 cars will take 22 hours, and 25 cars will take 46 hours, for a 28 hour average– If the failure rate was reduced to 10%, the average would drop to 24.4 hours, saving 3.6 hours with

no change in train speed

Now consider the situation where an outbound block leaves on two trains a day, evenly spaced every 12 hours

– Average dwell time will drop to 16 hours (10 hours processing, 6 hours of waiting time)– When a car misses its connection, instead of a 24 hour delay, it will now only be delayed 12 hours– In the above example, 75 cars will take 16 hours, and 25 will take 28 hours, for a 19 hour average,

saving 9 hours on average, with no change in train speed

The standing car count or car inventory for a yard has a direct impact on the fluidity of the yard – simply put, too many cars get in the way of operations

– Consider a yard processing 1,000 cars/day, with a 22 hour processing time– The above yard will have an average inventory of 917 cars– As the average time rises to 28 hours, this means 1,167 cars in inventory– At 36 hours, average inventory

reaches 1,500 cars– At $20/day in car hire costs, this

excess inventory over the 22 hourcase represents $11,667/day or$4.25 million/year

Number of Cars

Processing Time

Departures per Day

Average Wait Time

Base Yard Time

Failure Rate

Failure Yard Time

Average Yard Time

Average Standing Inventory

1000 10 1 12 22 25% 46 28.0 1,1671000 10 1 12 22 10% 46 24.4 1,0171000 10 2 6 16 25% 28 19.0 7921000 10 2 6 16 10% 28 17.2 7171000 10 1 12 22 50% 46 34.0 1,417


Improving Terminal Reliability

Many factors influence terminal reliability– Operating plans that are not well matched to traffic can result in annulments, extras, and

tonnage being left behind– Operating plans too tightly matched to traffic may not have sufficient capacity to handle volume

variations, also causing traffic to be left behind– Erratic inbound train arrivals can overwhelm the yard– Poor or inconsistent hump utilization– Delays in in-bound (or out-bound) train inspection– Availability of crews and locomotives– Early make-up of outbound trains can effectively reduce the time available to make connections– Poorly planned hump sequence, or train make-up sequence may adversely impact some

connections– Clogged receiving, departure, or bowl tracks– Disruptions due to car or locomotive failures– Blockages due to train or switch movements– Cherry-picking for hot cars– Yard maintenance downtime

Many of these issues can be mitigated through quality planning and management, and ensuring that the yard has the resources that it needs


Contents







Locomotive Fleet RequirementsAs discussed with rail cars, a host of activities consume substantial portions of locomotive time.

Railroads have reported that as little as 50-60% of road locomotive hours are consumed in physically hauling trains 1

The remaining time is consumed by a host of other time segments, including:

– Terminal dwell, including

- Servicing & Fueling

- Standing in terminals waiting for assignment (scheduled dwell, fluctuations in arrival and departure patterns, etc.)

- Idle time due to loco management assignment policies (locomotive matching, restricted service parameters, consist splitting/building)

- Delays due to inbound locomotive failures

– Maintenance activities (running repair, shop time)

– Repositioning/deadhead movements

Terminal Dwell, 30%

Hauling Trains, 55%

Mechanical, 8%

Repositioning/ Others

Locomotive Time Components(Indicative)

1 Highly dependent upon portion of locomotive fleet in unit train service, measurement parameters (e.g., when ‘on train hauling traffic’ measurements commence and terminate, average length of locomotive trip on same train, etc.


Locomotive Velocity – An incomplete measureMeasuring locomotive performance solely on the basis of line-of-road velocity ignores the impact of the range of activities that represent nearly half of total locomotive hours.

Increasing locomotive line-of-road velocity will result in improved locomotive utilization andproductivity – assuming all other factors are held constant

But, measuring performance based solely on line-of-road velocity overlooks a host of opportunities to improve locomotive utilization

– “Velocity” does not encompass the many non-train running factors impacting locomotive use

– “Velocity,” depending upon how measured, may record as “positive velocity” time spent repositioning locomotives to other locations/geographic areas to address imbalances or maintenance requirements

– “Velocity” typically does not adjust for non-productive excess HP/Ton

In fact, some railroads have found that focusing solely on line-of-road velocity induces behavior that actually reduces locomotive utilization

– Dispatching to maximize LOR velocity without regard for the implications upon locomotive standing time in terminals (“get the locomotive on the velocity clock”) or terminal/LOR congestion may reduce over-all locomotive productivity

– Assigning power in a manner that focuses on trains that will generate the highest “velocity,” not necessarily those trains that need to be protected to maintain service levels, may result in increased yard congestion and imbalances in locomotive maintenance work requirements

Some major railroads have replaced ‘velocity’ with more holistic measures to better measure locomotive utilization and productivity – measures such as GTM/HP day, or productive locomotive miles per day.


Locomotive ProductivityMany railroads have found focusing on dwell time reductions provides far higher returns than improvements in line-of-road velocity.

Studies have consistently found that achieving a 10% reduction in ‘standing time’ is both easier and faster to implement than a near-equivalent impact 10% improvement in locomotive line-of-road velocity

– Dwell reduction issues typically involve policy or process changes (frequently within a terminal) for which specific organizational units can be held accountable to implement and to deliver results

– Improving LOR velocity, on the other hand, typically not only entails a host of accountable organizational units (dispatch, road foremen, infrastructure, mechanical) in geographically dispersed locations, but can be highly dependent on external factors such as traffic volume fluctuations outside the direct control of the Operating Department

Studies have identified a range of actions to address standing time that have produced demonstrated results, including:

– Addressing policies that result in some locomotives experiencing very long terminal dwell times(in one major study it was found that over 50% of locomotive dwell time in excess of 10 hours was accounted for by locomotives with dwell time in excess of 24 hours)

– Improving locomotive inventory management practices (converting from responding to existing imbalances to proactive look-ahead processes)

– Minimizing the number of restrictions on the allocation of specific locomotives to specific trains, or due to mechanical defects

– Reducing processing time from arrive terminal to consist ready (in multiple studies average dwell time was reduced 3-5 hours per locomotive)

– Developing an effective short and medium term fleet sizing process that leveraged commercial volume projections to proactively manage the number of road locomotives available for revenue service

– Instituting revised train assignment and maintenance policies (revising policies on when to pull a locomotive from service for 92-day inspection in order to minimize dead-in-tow time, for example)


Contents







Manifest Traffic and Network FluidityProtecting the manifest network, and ensuring it operates smoothly is critical to network fluidity

The primary symptom in a poorly performing manifest network is clogged yards– This occurs at both the local and system yard level– At the system yard level, yards fill up, and can no longer receive additional trains– At the local level, yards fill up, and make switching operations difficult or sometimes impossible

Clogged yards end up impacting overall network performance in many ways:– Cause trains to be held out of terminals, blocking passing sidings, and slowing overall network

speeds and reducing line capacity– Cause trains to be by-passed to other terminals, increasing car and train miles, consuming

additional line capacity, and degrading yard performance further at the yards receiving the by-passed traffic

– Degradation of terminal performance slows manifest car velocity (and through spill over effects velocity of other lines of business), increasing total fleet requirements

– Erratic manifest train operations make crews and locomotives difficult to manage, increasing deadheads, idle time, and total crew and locomotive requirements

Not protecting the manifest network is the most common cause of clogged yards– The inability to depart trains is the primary cause of clogged yards– Terminals are continuous flow facilities, if you don’t depart trains, the whole system backs up– Trains are delayed in departing through lack of locomotives, crews, line capacity, and annulment

decisions – all symptomatic of a failure to protect the manifest network– When trains are then released on a “spasmodic” basis, too many cars are then sent to the

downstream yards, overwhelming them, and causing them to become clogged


“Scheduled” or “Precision” Railroading is Focused on Protecting the Manifest Network

The fundamental premise of scheduled or precision railroading is that by protecting the “plan” one can maximize asset utilization, network capacity, fluidity, and customer service – ultimately reducing costs

Statement by Tony Ingram (while VP-Operations at NS):

“With TOP, we created a more disciplined operating environment. Traffic moves by plan on scheduled trains, so our field operations managers are better able to allocate resources to meet customer needs. The way in which we plan work for yard and local crews is simplified as well. Yardmasters can process traffic through the yard facilities in a planned and consistent manner. Each terminal processes traffic to meet a connection standard. With more reliable on-time train performance, yard operations have become more consistent. This avoids the need to "replan" work to fit circumstances that vary each day.

“TOP gives us the means to improve our use of capacity across the rail network. For example, with more predictable train operations, terminals can allocate track space with a higher degree of certainty about the track space required, and the duration that the track is occupied. With more predictable train operations, we're able to assign locomotives from inbound train to outbound train more effectively. The end result is improved capacity utilization, lower operating costs and improved service reliability. With TOP's advantages, it's clear that the Norfolk Southern network is poised for growth. Our terminal facilities can handle additional volumes.”


“Scheduled” or “Precision” Railroading is Focused on Protecting the Manifest Network

To quote David Goode:

“We improved the throughput of our hump yards… Now that's building capacity without touching a piece of [hardware] -- that's just people, systems, and efficiency. The reason we resist giving you a percentage number on capacity is if we'd done that last year, we would have been wrong. We discovered we have a lot more capacity than we might have told you last year. I fully expect that as we continue to perfect and add systems, and we're doing that quarter after quarter, we're improving our operating systems, and as we do that, we're building train capacity – we’re building network capacity. We are excited about it and we're eager to see how far we can push that envelope.”

To quote Canadian National on Precision Railroading:

“With precision railroading – an evolution of CN’s scheduled railroading – the focus is clearly on the carload and the customer’s shipment, rather than on the train. That is what matters most to CN’scustomers – whether a train is on time or late is not of concern, but they do care that their shipments are on time.

“Within CN, precision railroading has made the Company more competitive and more reliable, and has improved its cost control and asset utilization efforts on the network and in its yards.”


Back-to-the-Future:The Freight Car Utilization Program Circa 1977

Overall the FRA/AAR Freight Car Utilization Program concluded:“…development of an effective operating / service plan and an organization geared towards

successfully executing that plan appear as key elements in fulfilling a top management commitment to improve car utilization and service reliability.”*

Specific 1977 FCUP conclusions:– Railroads generally do not know the quality of dock-to-dock service they provide to most customers– Most railroads do not give enough importance to service reliability and car utilization– Train operations, not total service performance, is generally emphasized– Car costs are often not associated with operating budgets– Operating department personnel do not make effective use of available data– Documenting the impact of service design changes on service is difficult, but important– When railroads focus on reliability and car utilization benefits accrue

- People respond to what they are measured and rewarded on– Railroads generally do not have clearly specified service standards for most movements– Railroads need to develop complete operating / service plans covering movement of both loads and

empties, and continually update it– Analytical tools can be used to develop more effective operating / service plans– Improved interline coordination is a necessity

* Freight Car Utilization and Railroad Reliability: Conclusions and Recommendations, October, 1977, Final Report


1977: To Succeed FCUP Said Railroads Must:

Have a commitment to service

Have a well defined, detailed operating / service plan

Ensure the feasibility of the plan

Build flexibility into the plan

Have a firm commitment to operating according to the plan

Have a control system to ensure plan execution

Understand what level of service it provides to its customers (or wants to provide)

Ensure coordinated decision making & awareness of impacts of decisions by all actors

Have the necessary data systems to support its goals

The road map does not appear to have changed in 30 years. While we have traveled a long way along this road, we clearly still have a long way to go

Asset Velocity

Documents

time terminal

time performance

idle time

terminal processing

yard time loaded

additional time components

cycle time manifest

trains storage time