In typical process supply chains, proper forecasting, planning and scheduling optimisation are essential to reduce costs, enhance operational service levels and minimise inventory. In the real world this may appear to be a highly complex problem, impossible to solve and full of potential pitfalls. However, advanced forecasting, planning and scheduling software exists that can reliably deliver optimised plans, and this software can easily be understood without a mathematics degree if the problem is broken down into the relevant layers and the basic principles of planning.
Examples of complex planning and scheduling problems abound in breweries, food processing, chemicals, pharmaceuticals and lubricant blending. These plants are characterised by the need to manufacture multiple products in batch or semi-batch processes with multiple production lines. The products need to be scheduled optimally to meet an anticipated demand pattern by the end consumer.
The problem is often made even more complex by a distribution network between the producing plant and the customer. Several physical constraints also apply, such as the need to blend liquids into tanks or take account of shelf life constraints. Alternative recipes and utility constraints, such as the availability of steam and cleaning utilities, can further compound the complexity of optimisation.
The levels of planning
Mathematically it is possible to define the planning problem quite precisely, and one can apply a number of computer-based mathematical techniques to obtain a production plan. However, selecting the right method, understanding and focusing on the factors that really matter and correctly interpreting the results of an optimisation is like playing chess. The rules can be described precisely, yet owing to the almost infinite number of alternatives, it takes a grandmaster to predictably make the right strategic decisions and play the right moves. In these situations experience does matter, and similarly planners with a good understanding of the business are a vital element.
However, business people also have an obligation to understand the planning process properly. Supply chain planning and scheduling is easily understood by decomposing the problem and looking separately at the different layers of planning:
1. At the highest planning level, demand forecasting seeks to predict overall demand for a particular product. This forecasting takes place with a long time horizon, often many seasons or years.
2. At the second intermediate planning level, network optimisation seeks to balance demand with production and distribution capacity across multiple plants and multiple stock points. The time horizon in this instance is typically months or weeks.
3. At the third, most detailed planning level, production scheduling seeks to interpret production requirements and make them applicable to a single plant by producing an optimised schedule that takes into account availability of physical equipment such as pumps, tanks and packaging lines. Here the time horizon is hours, shifts or batches.
4. Finally, the results are fed to the execution system, where production is actually executed on the physical plant. This is typically referred to as the MES or manufacturing execution system layer.
Forecasting vs optimisation
It is important to recognise that there is a fundamental difference between forecasting and optimisation, both in concept and in the mathematical tools and techniques used. Forecasting takes into account historical and other information to produce a realistic demand forecast for a family of products. The statistical methods used are usually highly advanced and finely tuned. The user of such tools does not need to be concerned with the actual statistical engine – the focus is on accurately predicting the future and accounting for abnormal events (marketing campaigns, the impact of major events such as the FIFA World Cup). The statistical engine will take historical patterns of demand, identify and ignore outliers (data that does not conform statistically to the pattern) and project this forward. In practice the software’s ability to interact with the user in a graphical and intuitive way is important.
Physical constraints
Optimisation, on the other hand, is concerned with optimising a result where there are several constraints on the variables. The variable being optimised is typically referred to as the objective function. Examples of an objective function could be total cost, made up of production costs, warehousing costs, transport costs and others.
The constraints used in optimisation are typically physical in nature, such as plant configuration or maximum production rate. However, constraints can also be soft – for example it is undesirable to increase the volume of road transport because of the long term impact on infrastructure and road maintenance. Soft constraints can be violated from time to time, whereas physical constraints cannot be violated.
The goal of supply chain optimisation is to derive first a feasible plan from the forecast, and then to refine this to an optimum plan. The word feasible in this context means that all constraints have been satisfied and the plan can be executed in the physical world. Optimum means that in addition to being feasible, the plan is the result of evaluating many alternative plans to derive one where the objective function has been optimised.
A typical planning process starts with demand forecasting which determines the likely consumption patterns for specific products. This demand plan is then cascaded to the planning system, which prepares a feasible and optimised production plan relating to overall targets, logistics and warehousing/stock parameters. Finally, the optimised plan is cascaded to scheduling systems to produce batch or shift plans at the detailed level.
It should be mentioned that no planning system would work without a feedback mechanism that compares actual performance against plan, and a process of refining and improving the plan on an ongoing basis.
Business decision-makers need only to understand the basics of the multilayer approach to planning and scheduling. While the detailed plans themselves are still prepared by the experts (the chess grandmasters in the company), with a fundamental understanding of the above principles, better business decisions can be taken, leading to reduced costs, improved service and lower inventory levels.
For more information contact Gavin Halse, ApplyIT, +27 (0)31 514 7300, [email protected], www.applyit.com
© Technews Publishing (Pty) Ltd | All Rights Reserved
printer friendly version