MP Modelling at Hoskyns

This article is based on a talk given by Ian Turner of Hoskyns Consulting (now part of Ernst & Young) to the Mathematical Programming Study Group.

Hoskyns

Hoskyns is recognised as having the largest and strongest commercial OR Group in the UK, and posssibly in Europe. Hoskyns' OR Group was established in 1992 with the takeover of British Coal's highly-regarded OR Executive. A further leap in size occurred with the acquisition of British Gas's OR Group at the end of 1994.

There are now over 100 OR consultants within Hoskyns Consulting. This is a skill centre with over 250 staff covering such activities as Business Process Re-engineering, IT Project Management and Technology Consulting as well as OR.

Within the OR Group, the full range of OR techniques is used, both hard and soft. The success of the Group may be gauged from the way in which it has flourished, acquiring new clients to sustain it as business from British Coal diminished and ultimately disappeared.

Ian Turner described four examples of Hoskyns' work involving Mathematical Programming.

Electricity Generation

In the run-up to privatization of the electricity industry, British Coal and its advisers needed to be able to assess the size of the market for coal and other fuels for electricity generation. Hoskyns built an optimization model of coal mines, ports and power stations which sought to mimic the way in which the privatized electricity industry was expected to operate.

Individual power stations were represented with their potential sources of fuel (coal, oil, gas), the costs of producing the fuel at the various sources, the costs of delivering the fuel from source to destination and the generating efficiency of the power station. The model also represented predicted demands for electricity around the country and the capacity constraints on moving electricity around the National Grid and through the interconnectors with Scotland and France.

It was expected that the generation system would minimize the costs of generating sufficient electricity to meet demand while satisfying capacity, plant availability and contractual constraints. Many runs were carried out to assess the likely demand for coal under a wide variety of scenarios. These helped British Coal and its advisers to assess which coal mines had a long-term future.

The model was built using the XPRESS-MP mathematical programming package with extensive data preparation and reporting routines implemented using a spreadsheet.

Supply Chain Optimization

Hoskyns' sister company, Gemini Consulting, was retained by the Board of a large Fast Moving Consumer Goods company to undertake a change management exercise. As part of this, Gemini Consulting looked at the supply chain from factory to shop and examined what benefits could be achieved by restructuring it.

Hoskyns built an optimization model which represented the manufacturing and distribution network and their costs. It sought to meet customer demand at minimum total cost. The model used Microsoft Access as a database for holding input data and results and had a Graphical User Interface written in Visual Basic.

With this framework, the consultants were able to experiment with changes to customer demand, supplying new countries, adjusting production sites and modifying the distribution network. The model showed the results in ways which were intuitive to management, thereby enabling its use to become an integral part of the consultancy process.

Gas Contract Management

The oil and gas companies which produce gas from the North Sea sell the gas under contracts which have a variety of terms. These cover such matters as the contracted annual take, maximum and minimum flow rates, whether take can be deferred, the price structure of the gas (there may be different prices for various tranches of gas) and penalties for operating outside specified conditions.

The purchaser of the gas has the problem of managing the portfolio of these contracts so as to minimise its costs while meeting demands. Demands are not known with certainty and depend on the severity of the winter. Some management of short-term demand is possible through the use of storage facilities but these cannot handle the sustained increase in demand which occurs in a severe winter.

Hoskyns has built an optimization model to help manage the portfolio of gas purchase contracts. It specifies the quantities of gas to take from the various contracts so as to minimize the long-term discounted cash flow of meeting demand. The model is primarily concerned with the demand for an average year but ensures that the demand for a severe year could be met. It can also "buy" extra storage capacity to ensure that a specified level of short-term peak demand can be met.

The model uses mixed integer programming and is built using XPRESS-MP with data input and reporting using Lotus 1-2-3 spreadsheets.

Many runs are carried out with different scenarios to assess how the contracts perform under a variety of conditions. Ultimately, the client makes the decisions as to the quantities of gas to take from each contract based on his assessment of the results of the model. Thus the model works as an aid to decision making rather than taking the decisions itself.

Vehicle Launch Planning

Hoskyns worked with a leading motor manufacturer to assist it in the planning process leading up to the launch of a new range of cars. Such a launch is critical to the company's success and needs to be coordinated across Europe, both in terms of the manufacture of the cars and in terms of having the correct models in dealers' showrooms at the right time for the launch.

Modern cars are manufactured in an enormous number of variants and the demand for these varies from one market to another. Estimates are produced of demand, but in the launch period it is more important to make full use of production capacity and deliver cars to markets at the correct time than to supply precisely the correct variants.

Hoskyns built an LP model which represents the manufacturing facilities, the demand for vehicles in various markets and the distribution system. It seeks to schedule production to meet the demand as nearly as possible while making full use of manufacturing facilities and satisfying other "soft" objectives. These include maintaining stable production and supply patterns (i.e. using the standard factory to supply one market if at all possible) and sharing the pain of minor production limitations.

The model has been built using XPRESS-MP with Microsoft Access providing the data entry, reporting and database facilities. It has now been used on several model launches with a Hoskyns consultant acting as interpreter between the model and management. He or she translates management's aims into the model's terms, runs the model several times and then talks through the resulting reports.

Common Threads

These models have all been built where the client has had an explicit need for a model. They have generally been part of a wider consultancy effort, either in Operational Research or in Management Consultancy. The models have been designed either to be used by the client directly or to be used by Hoskyns consultants working very closely with the client. As a result of heavy client involvement, they have all evolved considerably during use.

Much effort has been expended on the data handling and reporting systems. The more recently-developed models have used Microsoft Access with considerable success and, in one case, Visual Basic as well. In this way the models are tending to hide their LP origins and are becoming part of more general decision support systems.

Related articles include Aluminimum Smelter Benefits from MP Consultancy and Prize-Winning Planning at Harris Semiconductors. To find other articles, refer to the MP in Action page.

Comments