Вы находитесь на странице: 1из 2

Zachary Garten

Natalie Holt
Andrew Perry
Improving the Distribution of Industrial Gases
In 1940, Air Products and Chemicals revolutionized the industrial gases industry
by building facilities adjacent to large volume users so that gases could be piped directly.
Since then, Air Products has led the industry towards the goal of maximizing efficiency
by heavily investing in optimization tools that reduce distribution costs. Industrial Gas is
the most innovative division of the company that has developed computer applications to
improve productivity for fleet transportation. In order to further their competitive edge, Air
Products began to automate their delivery scheduling in 1975. The 23 plants that
manufacture liquid oxygen and nitrogen also serve as supply depots that distribute to
about 3,500 industrial users and hospitals. Distribution management has to account for
this large number of customers in addition to: usage rates, hours of operation, mileage,
truck capacity, drivers, and vehicle related costs. In 1980, Air Products introduced a
computerized vehicle scheduling program to more efficiently address the complexities of
the problem. The module is now used daily at each depot to produce a detailed two to
five day schedule that can be easily updated with customer requirements as they are
received. Six files are used in the system including: Customer, Resource, Cost, Time
and Distance, Mileage, and Schedule. The data within these six files are used in the
scheduling module known as ROVER (Real-time Optimization for Vehicle Routing) which
solves through a mixed integer program of up to 800,000 variables and 200,000
constants.
Dealing with such a complex problem is challenging, necessitating assumptions
to simplify and hard work to overcome barriers. Development was difficult, as mentioned
Air Products intended to optimize scheduling costs, minimizing cost equations dealing
with 800,000 variables and 200,000 constraints. The sheer complexity of the problem
requires a high level of precision by everyone developing and using the system.
Assumptions about the data were made, such as hard constraints like safety stock level,
as well as assumptions about totally control and freedom to schedule routes. Costs were
assumed to be transportation costs, prepping cost, vehicle costs and the like. At the time
of this project, computer technology was expensive, so the cost benefit of outfitting
depots with computers was tested. Also to be considered, the challenge of actually
computing optimal answers. Computing power was a constraint for developers, and thus
developers had to knapsack delivery routes in and solve the problem in pieces. The
Lagrangian optimal solution had to be relaxed to accommodate the complex problem,
and it is assumed that the solution is close enough to optimal. The program would not
work with routes with many stops, although this was uncommon. It was assumed routes
would include two to four customers. It is also assumed that routes can be scrapped if
they are found to be not interesting or profitable enough. Implementation was its own
challenge, getting schedulers to utilize their program that took 24 man-years of effort to
put together. Schedulers werent the most knowledgeable with computer technology, so
engendering trust with the system by the schedulers took focus and planning to
successfully implement their project. Working through the numerous bugs was also a
challenge with implementation. Schedulers were used to certain elements not present in
the program, such as soft constraints that were more common than the hard constraints
offered. Also, small changes in the constraints could have a drastic impact on the

optimal route, which was different than what schedulers had experienced in the past.
Schedulers were also wary of accepting a technology that could be perceived as putting
their own job in jeopardy. The high risk- high reward nature of the project compounded
with the pressure of the grim economic picture made this a critical project to the future
success of Air Products. Air Products overcame these challenges endlessly working with
stakeholders, making sure that the ultimate end user understands the benefits and
utilizes the tool for success.
In terms of implications that the program has had, the obvious implication of the
ROVER program is that it allows trucks to be routed in such as way that they have to
travel less miles overall. This allows companies using the product to save money and
therefore cut their cost of shipping so that they can attract more clients. This implication
can be directly seen in the case of the Wharton location where the average miles per
gallon delivered was decreased by over nine percent. This reduction in cost of shipping
is predicted to save a company around 1.6 million dollars annually. Another implication of
the program was that it will decrease the number of trucks that a company has to buy,
since their current fleet is running more efficiently, which can save a company over 3
million dollars annually. As well once the infrastructure of the ROVER is put in place
there are only small costs to expanding the network of coverage for the ROVER. The
real success of the ROVER program is that once the initial installation of the program is
done it is inexpensive to expand it, and as time goes on it becomes more and more
effective. The schedulers are able to make decisions more quickly using the program,
making their lives easier. The company executive are excited that the program is saving
them money on capital investments and day to day costs. The companies receiving the
gasses are happy since the cost of the shipping to their facility is decreased. Therefore
ROVER satisfies all major stakeholders in the shipping process, with the only entity
possibly upset, the truck drivers, since they are paid by the hour and now have less time
on the road since they are routed more effectively.

Вам также может понравиться