|
ABSTRACTS
Kraft, E. R. (1982) Jam Capacity of Single
Track Rail Lines, Proceedings, 23rd Annual Meeting of the Transportation
Research Forum 23 (1) 461-471.
Jam Capacity is defined as the maximum short-term rate at which a single
track rail line is capable of moving traffic after "Jam" conditions
have developed. It is well known that highways have two distinct modes of
operation - free flow and jam - under which the same traffic volume can be
moved and this is true for rail lines as well. Jam conditions develop when two
fleets of trains meet enroute and one or both of the fleets must be broken
up. The last trains in the fleet must take long delays because sidings ahead
of them are filled up. A methodology is proposed for estimating the jam
capacity of a rail line and for reducing that to a practically attainable
value. An example application of this methodology to a real world rail line
is given.
Kraft, E. R. (1987) A Branch and Bound
Procedure for Optimal Train Dispatching, Journal of the Transportation
Research Forum 28 (1) 263-276.
The paper outlines major design goals, limitations and capabilities of
train dispatcher assist systems and the simulation techniques upon which they
are based. A probability model of train delay is derived to show how
dispatching decisions can be made to minimize the priority-weighted sum of
train delays. A deterministic branch and bound algorithm is presented and its
performance compared with this "local optimization" technique.
Simulation testing indicates delay can be reduced by as much as 55% under
heavy traffic conditions. Using a random number generator, running time
variability was introduced into the simulation. Test results indicate the
procedure is not overly sensitive to random speed fluctuation. As running
time variability increases, delay savings remain almost constant in absolute
difference. On a percentage basis, savings decline because the base level of
delay increases.
Kraft, E. R. (1998) A Reservations-Based Railway
Network Operations Management System, Ph. D. Dissertation, Department of
Systems, University of Pennsylvania, Philadelphia, PA.
A
shipment scheduling and operational control method is developed and tested,
to help railroads become more competitive for high revenue, service sensitive
freight. A bid-price based, profit maximizing revenue management approach is
proposed to allow a rail carrier to develop achievable and market sensitive
quotations of delivery time for new shipments calling in. Bid prices are
derived using a subgradient step size algorithm; a modified shortest path
procedure is used to solve the decomposed subproblems.
Once
the service quotation has been developed, a deterministic, cost minimizing,
multicommodity network flow model, including train capacity and integral flow
constraints is solved to manage shipments moving on the railroad in real
time. This model dynamically reroutes shipments to take advantage of all
available train capacity in the network, while still meeting the committed
delivery times on priority shipments. A customized dual ascent procedure,
using a tabu search approach as an anti-cycling mechanism, adapts the
previous solution any time new information is received.
Both
procedures have been integrated into a rolling horizon simulation model.
Simulation results indicate up to a 10 point improvement in railway operating
ratio may be achievable through implementation of this shipment management
strategy.
Keywords: Railroad, Railway, Revenue Management, Yield Management,
Car Scheduling, Shipment Scheduling, Bid Price, Optimization, Simulation,
Service Monitoring, Subgradient Algorithm, Tabu Search, Service Management
Home
Last Updated: January 11,
2001
|
