Resources are limited in any organisation, regardless of industry sector. Whether the limitation relates to people, equipment, facilities, or any combination of these, it restricts the ability of the business to operate as well as it would wish. The success, or otherwise, of an organisation in making the best use of its finite resources has a major impact on perhaps its largest controllable costs.
Consequently, a great deal of money and effort goes into trying to maximise resource cost-effectiveness, both in identifying the right business processes and in choosing computer-based systems to assist with resource planning and scheduling. Unfortunately, this is often misdirected, resulting in no improvement in resource utilisation and sometimes even in a degradation in performance.
According to PA Consulting Group, a leading management, systems and technology consulting company, common reasons for organisations failing to meet their objectives, are that they are attempting to make decisions based on data of the wrong type or level of detail, the decision support tools used are inadequate or inappropriate, or they are not aware of what is achievable.
Corporate information can be used to reap benefits from limited resources. However, two fundamental principles must be honoured - those of 'the planning horizon' and 'fitness for purpose'. PA's work at Nissan's Sunderland plant demonstrates the application of these principles in an automotive manufacturing environment, see after article.
The planning horizon
The planning process covers a time horizon that stretches from months or years away down to the actual day of the event, or even beyond. The planning objectives and questions to be answered at each point in this planning horizon may be very different.
The focus of interest at the different levels moves gradually from an emphasis on 'what if' investigation to operational management as the time of the event approaches. Interest at each level is typically:
- Strategic planning – a company may be considering such questions as the resource implications of providing a new type of business, evaluating the cost-effectiveness of possible major investments and the impact of increasing multi-skilling or greater flexibility of resources.
- Capacity planning – the issue here is more about understanding whether there will be enough resources to meet the anticipated future need to enable any necessary steps to be taken.
- Scheduling – the focus changes in making best use of the resources available to meet known demand within reasonable bounds of accuracy. Concentration is usually on 'logical' rather than 'actual' resources; that is, on the need for a given number of resources of a particular type, not on which specific individual resource will be assigned to meet which specific need.
- Rostering – actual people or specific pieces of equipment are identified for specified duties or tasks, and this usually needs much greater detail than the scheduling step.
- Reacting – concerns the best way to handle short-term, unexpected events. This could range from minor reallocation of resource if a specific task is overrunning, through to a drastic rewrite of plans following a major incident.
- Recording – this is not included in all planning processes. It relates to what was actually done, rather than what was planned, and is most relevant where tracking is important or where the rosters are used as a basis for payment.
At the strategic and capacity planning level, data and results tend to be required at an aggregate level. This is because the focus is on determining an overall direction and it is not necessary to know the specifics of what particular resources will be used to meet what need. In contrast, at the other end of the time horizon the data and results need to be precise, since specific resources are being allocated.
The level of detail implied above has implications both for planning tools and for data needed to support them. The focus at the strategic end of the time horizon will be on flexibility to investigate a wide range of potential options, whereas, at the operational end of the horizon, it will be on dealing with the detail in the most appropriate and efficient means possible. Similarly, the most appropriate technical planning solution is likely to be different at each point of the horizon.
Fitness for purpose
One of the most challenging aspects of specifying a planning solution is ensuring that the supporting IT system is precisely appropriate for the purposes intended, so that that it can be designed or chosen with the business objectives at the fore from the outset. Whilst this appears obvious, it is surprising how rarely it is even considered. All too often, planning solutions are data driven, not business needs driven. The most common approach is bottom-up:
- establish what relevant data is collected
- describe all the physical processes in the operation
- specify the system requirements in terms of using all the data and representing all the processes.
This rarely leads to the right solution and is often destined to fail. Failure comes either from the business needs being completely lost, or in a level of detail required from the data that is unsustainable or inappropriate for the questions being asked.
In contrast, PA's approach is to measure all potential elements of a planning solution against a series of business-focused questions, such as:
- What decisions will be based on the results of the system?
- What are the business requirements of each element in the planning process and what business benefits would be brought by meeting them?
- Can the necessary supporting business processes be put in place to ensure that the planning solution is practical and the potential benefits achievable?
- Can the data be provided and maintained at the necessary level of detail, or are there inherent uncertainties that place a practical constraint on the level of detail at which it is sensible to plan?
- Will the users support the proposed solution?
- Is there a clear migration path from the present through to the final solution?
In PA's experience, the user and business process issues are commonly ignored. Yet, unless people are committed to the solution and the mechanism is in place to make it work in the business, the solution will fail, no matter how theoretically ideal it is. This can mean making compromises: it is far better to have a more modest solution that actually works and delivers some benefits than one that is ideal in principle, but for whatever reason cannot be made to work in practice.
Finding the right technical solution(s)
There are a series of fundamentally different technical approaches to planning and scheduling, both in terms of the mathematical techniques available and how they are presented to users in a 'product'.
The diversity of techniques is ever increasing and ranges from very simple, well-understood methods, through to leading-edge techniques that have only recently become available for industrial applications. Each has its relative strengths and weaknesses, and many problems can be addressed by more than one method. However, in some cases the choice of an inappropriate method could either lead to failure or to a solution that is far more costly to develop than necessary. When choosing a technique, it is important to seek advice from experienced practitioners, who know exactly what is required to make each type of solution work, and are aware of the possible strengths and weaknesses.
In terms of the representation of planning techniques within products, there are several fundamental types:
- planning functionality within enterprise resource planning (ERP) systems
- single-purpose planning packages
- bespoke developments
- bespoke applications built round a pre-developed commercial 'engine'.
Some products are geared to the purely operational, others to answering strategic questions. There is often a 'grey area' where most products are designed primarily for one part of the planning horizon, but also attempt to provide functionality for other parts. However, it is highly unlikely that any single product has the ability to offer the best possible solution right across the planning horizon - beware the salesman offering a product where 'one solution fits all'.
ERP systems have always had some scheduling capability, albeit usually very limited. Some are now attempting to break into the high-complexity planning and scheduling arena through the addition of advanced planning modules that connect with the underlying ERP data engine. The potential benefits are in integration with the operational data, removing the need for potentially costly interfacing. The risks are in the level of detail at which the data is often represented and, more importantly, in their 'black box' nature and the impression often given that they are simply 'plug and play' implementations, when nothing could be further from reality.
The benefit of single purpose packages is that, if they meet the business need, they will usually be the cheapest option. However, if there are any complexities or idiosyncrasies in an operation, packages rarely accommodate them. Similarly, the merits of purely bespoke solutions are well understood. The trade-off is between the tailored functionality and the risk and cost of a bespoke development project.
An emerging and increasingly popular middle ground between package and bespoke approaches is the use of commercial 'engines', designed to solve the specific types of mathematical problems found in the planning and scheduling of finite resources. A specific application is built by describing the problem in a way that the 'engine' can understand, offering the benefit of tailored functionality without all the risk of fully bespoke development. The potential difficulties of this approach are that it requires very detailed understanding and skills that are relatively rare.
Delivering the business benefits
This article has described some of the fundamentals that must be addressed when considering the development of planning and scheduling systems. It is possible to deliver significant benefit and achieve something never before possible; but, equally, it is possible to make major mistakes through well-intentioned, but misguided, action. In summary, the difference between success and failure can be made by:
- driving the solution from the business need, not beginning with a bottom-up assessment of all the data and processes present
- choosing a solution that is fit for purpose, both technically and in terms of the planning horizon
- making sure that the chosen approach can be supported by availability of the necessary data and is consistent with the level of uncertainty in the data.
The need to make the best possible use of finite resources will continue to be high on the agenda for many organisations. Indeed, its importance is likely to increase due to such pressures as the quicker response requirements of e-business and ever increasing customer service expectations. The mathematical and technological approaches are now available to solve almost any finite resource planning problem. The challenge is to ensure that the approach chosen is appropriate and, most importantly, fit for purpose.
Tony Lewins is a consultant at PA Consulting Group and specialises in finite resource optimisation.
Finite resource planning in practice - case study
Nissan Motor Manufacturing (UK) Ltd is a global car manufacturer that produces around 334,000 cars each year at its Sunderland plant in Northern England. The plant is recognised as the most efficient in Europe and the eleventh most efficient in the world. As a result of the plant's production excellence and efficiency, Nissan decided that the Sunderland plant would produce the new Almera model for the European market, whilst also maintaining production of the Micra and Primera models.
However, Nissan Sunderland was then faced with the complex issue of meeting the production requirements for the additional model, as cost-effectively as possible, with only two main production lines and very tight deadlines. PA Consulting Group was chosen to work in close co-operation with Nissan and ILOG, the market leading supplier of complex scheduling software, to develop a new scheduling system.
Developing the scheduling system in partnership
To deliver the third model, Nissan needed to increase output efficiency even further, making optimal use of existing facilities through tighter and more accurate planning. This could only be achieved if over 10,000 vehicles per week were built in the best possible sequence whilst satisfying a large number of operational constraints. Each of the many combinations of model specifications and colours can have a different workload impact. This affects all of the production steps, covering initial body build, paint shop and fitting out in the trim and chassis lines. Great care must be taken to balance the flow of work to avoid bottlenecks that would slow down the production line. For example, a sequence that seems ideal for the body shop may prove to be totally impractical later in the process.
PA combined business insight, expert systems development capability, detailed knowledge of scheduling techniques and mathematical expertise to develop a viable solution for the scheduling of the third vehicle model. This Body Build scheduling system generates an end-to-end sequence for a whole week's production for the Sunderland plant, and uses PA's own algorithmic methods together with advanced scheduling tools provided by ILOG.
The project was challenging because of the complexity involved, and because the PA team was developing a system to model a physical production operation that did not yet exist. This meant that there were many unknowns whose solutions could only evolve over time. It was vital to support the Nissan team of engineers to make operational decisions about the uncertainties and to accommodate these within the scheduling system.
PA's approach was to work in close collaboration with Nissan's production planning staff and ILOG. Nissan's production schedulers were an integral part of the team, ensuring that every factor affecting the production process was incorporated into the system. In the initial stages of the project, the PA team interviewed over 50 Nissan production management staff to identify over 2,500 constraints and preferences that were relevant to the scheduling model.
The system went live in August 1999 and has scheduled the production at the plant ever since. It is believed to be the first tool of its type in the world that is capable of producing an end-to-end production sequence for three car models whilst accommodating cross-overs in the plant and allowing multiple models on a single production line. The plant has realised an increase of 30% in the potential throughput without any major additional plant investment.
Benefits of the new scheduling system include:
- all production scheduling is done without the need for resequencing vehicles while they are in 'storage buffers' between major sections of the plant – storage buffers are used by many car manufacturers to help overcome scheduling problems
- the system also supports strategic 'what if' investigation so that, for example, the impact of potential changes to operational rules or constraints within the plant can be fully investigated
- schedule adherence rose from 3% pre-system to 85% after the system's implementation
- schedule results can be produced in minutes instead of days.
