Capacity Management

This article describes our approach to Capacity Planning and Control encompassing, Goldratt and his "Theory of Constraints", Rough Cut Capacity Planning (RCCP), "Participative Master Production Scheduling (MPS)", Lean Manufacturing and Supply Chains and "Pull" systems. It also discusses how to avoid spending lots of money on Advanced Planning and Scheduling (APS), Work flow, BPM or MRP2 systems, Finite material and capacity planning (including OPT, and PERT networks)

Links to other best practices and training at bottom of page.

There are two components of capacity management:

1. Capacity Planning (creating sufficient, flexible, capable, capacity & a valid, best, "Do-able", resilient, plan, to accommodate demand)
2. Capacity Control (ensuring the plan is met by managing resources)

Without capacity (and materials) to meet the demand, the plan cannot be valid.

A. Capacity Planning

There are, in a typical business, four levels where capacity planning (& control) is required as shown below. At each of these levels there may be a one to many relationship with the level below. There are certainly differences in both detail required to satisfy this level and planning horizon, which in the case of strategic planning at level one, product groups are being forecast with an horizon of many years. At level four, when you are managing an individual resource, you are dealing with detailed operating instructions and horizons of seconds:

Taking each of these levels in turn:

1. Strategic Capacity Management (business planning) includes capacity planning relating to:
   • Strategies for improving output or responsiveness
   • Core competence:
     • Providing competitive performance & meeting Critical Success Factors (CSF's) for the business
     • Resources capability (but avoiding the use of Overall Equipment Effectiveness (O.E.E.)!
     • Make vs. Buy
   • Capacity (line) balancing
   • Simplifying the process & creating value streams (Reducing the need for scheduling)
   • Replacement theory (Repair or replace?)
   • Constraints (We discuss three types of hard capacity constraint and how to avoid or minimise the impact of them using a number of tools & techniques such as "capacity exchange curves", "the capacity battery principle", "redundant methods", and "reducing the impact of forecast error", in our S04: Strategic Capacity Management Training Course). These constraints are characterised below:
     1. Hard Ceilings, where it is extremely difficult to add capacity e.g. expensive plant or equipment working at full capacity, or a scarce skill. (Also see Participative Sales & Operations Planning.)
     2. Hard Floors, where fixed costs in the business are high and is is difficult to remove them, due expensive equipment or scarce skills employed and underemployment in the lean times creates financial difficulties
     3. Hard Walls, (sometimes referred to as band width, i.e. the inability to accommodate simultaneous variety, even though there is sufficient overall capacity), where plant or people are insufficiently flexible to accommodate the range of varieties or changes in mix in particular but also volume (no multi-purpose / quick change equipment, or lack of versatility or mobility in the people (See Agile Manufacturing.)
   • Creating resilience & consistency of service:
     • Managing volume change & variety
     • Devising strategies to manage seasonal demand
     • Risk / Business continuity (See Process FMEA)
   • Turning the business plan into a capacity plan
   • Justifying capital expenditure, or producing profit forecasts
2. Development, Sales and Operations Management (Management of the demands on the business and the gross capacity to meet it) (The Sales &  Operations part of this process is mainly covered in a separate article "Participative Sales and Operations Planning".)(See below.)
   • Traditionally development planning has been excluded from the Sales & Operations Planning process, but we have found that not only does development vitally influence the timing of sales & operations plans, but it also consumes significant amounts of resource, so needs to be integrated into this process. However in some businesses with stable products it is possible to make simplifying assumptions about the amounts of resource needed for development.
   • It is important to distinguish between constraint types when capacity planning. Hard ceilings in particular need to be included in high level Master Production Scheduling, but walls and floors may also be significant at this level.
   • Tools, techniques & methods to manage this level include Participative Master Production Scheduling, "TAKT time" & "Rough Cut Capacity Planning"
   • There are also 11 separate degrees of sophistication which can be applied to capacity planning, ranging from the crude approximations of input  / output control below, to artificial intelligence / heuristics. Generally the more complex the situation, the more sophistication needed. This will be covered in a future article.
   • Whilst often forgotten in the complications of the budgeting process, capacity has to cope with peak demand not average demand in order to satisfy individual customer needs. The difficulty is to satisfy peak demand constrained by a budget containing average costs. This often leads to capacity lagging behind demand in an upturn even if the demand is accurately predicted. Conversely even if a downturn in demand is accurately predicted, the backward looking financial control systems do not create cost reduction tension until too late. This is a problem which management accounting has not yet properly addressed, but fortunately most general managers have a weather eye on the order book and sales pipeline to try to keep costs and income in line. However this check should be a routine part of the Sales & Operations Planning process at level 2.
   • The "Theory of Constraints" (TOC) penned by Goldratt & Fox in their book "The Goal" argued that the capacity of the supply chain system was governed by the capacity of its weakest link (the bottleneck) and that overproduction in other areas would simply produce unwanted inventory. Therefore high level control needs to be exercised to avoid local optimisation. Advanced Planning and Scheduling (APS) begins to miss the point that if sophisticated models are needed, perhaps the manufacturing system is too complex and should be simplified. ("Period Batch Control" and "Production Flow Analysis for Planning Group Technology" by the late Jack Burbidge). This thinking has led to the popularity (but not yet wide-scale adoption) of the concept of "lean manufacturing" and "lean supply chains", which is covered in more detail elsewhere on the site. (See below.)
   • Capacity planning can be simplified by creating representative models of the real world using a capacity model based on critical or bottleneck resource availability and by interpreting the demand on that resource alone to determine the overall likely output. This technique is called "Rough Cut Capacity Planning" and provides a rough check that demand and capacity are in balance. This process was originally envisaged to be the role of an individual called a "Master Production Scheduler" ("Manufacturing Planning & Control Systems" by Volman, Berry and Whybark), who would present the output plan to operations to produce. This concept was inherently flawed in that only that person owned the plan. If this check (which can usually be done on spreadsheets) is in place, a process can then be built around this to involve the stakeholders in a planning process. This technique "Participative Master Production Scheduling" (PMPS) ensures input from the participants to the plan and thereby commitment by those stakeholders to its achievement.
   • By its very nature Master Production Scheduling (MPS) attempts to smooth demand to produce a stable operations plan and thereby either, produces items earlier than needed, or increases some batch sizes beyond immediate requirements in anticipation of future requirements, both of which may be using up capacity required for immediate customer requirements, and produce unnecessary inventory. These practices have to be viewed as sub-optimal to producing what the customer wants, when they want it. However in many businesses demand is influenced by seasonal or other factors which make stable demand impossible. Or many processes or suppliers' processes are inflexible, which make changeovers from one job to the next time consuming. The danger is that these constraints may be viewed as immovable objects when in fact they can often be removed or alleviated. (See Agile Manufacturing.)
   • When capacity is approached by the demand, lead-times start to increase disproportionately. In the figure below the first period is full and there is some capacity available in the second period to accept further orders (Available To Promise / ATP). So currently the lead-time is 2 periods. If an order is taken which fills this capacity the lead-time is now 3 periods. It is interesting to note that the sales process from which the promises are derived often ignores the dynamism of this relationship. This results in unachievable promises being made to customers, which leads to frustration, expediting and a breakdown of the planning system. Secondly when a new product is launched the lead-times are in fact longer than they will be later, because later, the learning curve will have been climbed, supplier relationships established, snags removed from the design etc. It is also interesting to note that this reduced lead-time is often not later, reflected in the sales process.

     

      

3. Workflow Management / Scheduling (Scheduling of individual functions, cells or process areas)

   • In the mid 1970's the commercial availability of computers also spawned capacity planning tools, whose models were very sophisticated even by comparison with today's systems. It is mathematically possible to create a comprehensive model of the manufacturing or supply chain processes run on powerful computers, which use a variety of optimising techniques, in recently created "Advanced Scheduling Systems", in order to schedule work. (Previous Technique of the Week 020: "Close Scheduling" provides an introduction to scheduling.)

   • Also MRP2 Systems took the MRP1 plan and scheduled operations to create a "work-to" list at operational level in the early 1980's.

     However it is difficult to justify the additional cost and administration that these systems require, if:

     • Less sophisticated processes such as an effective master production scheduling process supported by a rough-cut capacity model is implemented first.
     • Level scheduling (See Previous Best Practice of the Week 005: "Level Scheduling") based on good sequencing is employed.
     • Work in process is kept small by the use of Input / Output Control as shown below or "Pull / Kanban" systems (see Materials Management & Stock Control).
     • Simple loading / planning boards are adequate. (See Previous Technique of the Week 017: "Loading Boards")

     Input / Output Control

     

     • Actual input should not normally exceed actual output to avoid work in process building up and complicating the process
     • Planned and actual output should be equal (On Time In Full) (see Previous Best Practice of the Week 046: "OTIF Measuring On-time Delivery")
     • Planned input should not normally exceed planned output unless you are going through a period of priming, ramp up, or ramp down

     There are a number of mechanisms which can be employed to manage workflow at level 3 before considering scheduling systems including:

     • Pull Systems (see Materials Management & Stock Control) or Kanban systems
     • Input / Output control / TAKT control
     • Out Tray Management (See Previous Best Practice of the Week 035: "Out Tray Management")
     • Queue management (Future article)
     • Measuring Operational Effectiveness (OTIF)

Finite material and capacity planning (including OPT, and Advanced Scheduling Systems, and PERT networks)

There is no doubt that mathematical approaches to scheduling are both valid and precise. You should bear in mind though that you can be precisely wrong!, and there are some practical problems. Businesses are complex, uncertain places. So to be accurate the mathematical models must reflect the complexity and statistical uncertainty of reality. This leads to a number of practical problems.

1. It requires a specialist to run it.
2. Potentially very large computer models are required, which are unstable if there is uncertainty e.g. Absenteeism, quality problems, process breakdowns etc.
3. It is very difficult to understand why the "work to" list says what it says, and the plan is imposed not agreed.
4. Administration of the data is very high which means:
   1. It is inefficient
   2. It becomes inaccurate very easily

These systems were widely used before the implementation of product-focused processes and Kanban systems, which have proved far simpler and in many cases superior.

"Drag & Drop" electronic loading boards have some utility in resolving scheduling problems where a manual loading board has reached its limits, but they suffer the same disadvantages above.

There are a number of issues relating to the maintenance of valid computer capacity models:

• The continuity of support
• The validity of models with uncertain processes
• The ownership of the resultant plan
• The understanding of the resultant plan
• The lack of stakeholder participation in the process

There is a further issue also relating to the documentation of the process, which can become a barrier to change. For example in the pharmaceuticals industry, process control documentation is of paramount importance as a quality assurance and control mechanism. However the documentation of the capacity planning parameters can create significant administration.

You need to take a staged approach to implementing sophisticated scheduling tools of any description. What you need to do is:

• Remove complexity from your operation by Organisational Redesign techniques
• Remove variability from your processes
• Increase the Agility of your processes
• Evaluate advanced systems with a view to redesigning your operational planning and scheduling
• Redesign your planning process to take into account the remaining variables
• Select the degree of sophistication you now require to deal with the remaining complexity
• And then postpone spending the £100,000 or more on the computer-scheduling tool by implementing the simple methods first and then justifying the next level of sophistication a stage at a time

These arguments are still valid in the context of infinite capacity plans contained in many MRPII systems. We describe this process in detail in our courses M02 Advanced Scheduling Systems & M23 Capacity Management.

4. Process Management (E.g. individual settings, speeds, feeds, skills, set / make ready times etc.)

   At level four we have been involved in some interesting & lively debates about:

   • What is the best method (running speeds, feeds, process settings, least waste, shortest lead-time etc.). This is where Taguchi methods (Design Of Experiments) is particularly useful. (In one recent example we showed that by reducing conveyor speed, more throughput could be achieved.)
   • Processing frequency / batch size, & in particular why work must be done in large batches, which "Level Scheduling" and "SMED" (below) specifically address.
   • Identifying and then driving skills development (versatility / mobility) using skills matrices (See Previous Best Practice of the Week 006: "Scarce Skills Management")
   • The worst case changeover time we have encountered so far is 9 hours for a single "efficient" process. This was accompanied by significant run down and run up materials losses, inhibiting flexibility significantly. The best result we have seen so far is 2 minutes for changeover of a large, vehicle body panel, press die. It is this area that "SMED" (Single Minute Exchange of Die) techniques ("A revolution in Manufacturing: The SMED System" by Shigeo Shingo) has much to offer.
   • How to measure process effectiveness & what are the real influences on process effectiveness. (See below)

   We deal with all of these issues in courses M11 Simple Ways to Maximise Output & Workflow & OM02 Managing & Improving Individual Skills & Overall Skill Levels

B. Capacity Control

Firstly we believe that the 3 dimensional approach of Overall Equipment Effectiveness (OEE) is inadequate, and the six big losses of OEE incomplete. In fact we have identified 21 so far, each of which needs to be resolved individually. (See Previous Technique of the Week T007: "CARAT" (Process effectiveness measurement, or why OEE / OME is for the birds)")

Secondly much is made of the process of capacity planning and in particular in the availability of sophisticated re-planning tools. A university professor recently stated that all of their post graduate research projects in manufacturing systems engineering were dedicated to seeking the holy grail of the ideal scheduling algorithm. However if as much attention was paid to meeting the plan instead of constantly changing it we think the process would be significantly more productive and also constantly improving.

Thirdly much is also made of "sweating the assets" or "maximising productivity". In fact there is only one asset in your business that needs to be operationally sweated & that is the bottleneck. Often this bottleneck is a service area or sometimes, if the order book is low, the sales department. (See Malpractice M006: Hitting the numbers.) Sweating a non-bottleneck will produce unwanted output! To  illustrate of this point, answer the following question: How much output should operations ideally produce if the order book is empty? The answer of course is zero. So why are operations measured on maximising output? The key question is how can we de-bottleneck, perhaps by reassigning underutilised resources . We have developed a new way of defining resource capability, de-bottlenecking and getting more output from them which we teach in M05 Simple Capacity Planning & Control.

Capacity control operates at all four levels:

1. Business Planning (where typically budgetary type controls operate)
2. Sales and Operations / Master Production Schedule, where the overall plan is measured and performance against the plan analysed and actions taken to bring output into line with demand.
3. Workflow which is typically short term, which can be minutes, hours or days depending on the lead-time, where short term actions are taken to bring the plan and achievement into line
4. Process Management conformance to requirements. (E.g. Why everyone is not using the same "best" method.)

The difference is really the level in the organisation where the decisions need to be taken, which depends on the impact. This tends to be strategic & long term at business level, but tactical & short term where the horizon also tends to be shorter for local decision making.

The timeliness of control is a key element, which can be illustrated as follows:

In the left hand diagram if action is taken in a timely way to respond to a required increase in output, the deviation can be corrected fairly easily. If the action is delayed the shortfall (in this case) in the right hand diagram has accumulated and recovery is much more difficult. In one case we were involved in, a 28 week recovery plan was needed to remove a 3 week backlog.

There needs to be a mechanism to exercise the control involving the stakeholders. This implies a meetings structure to discuss the issues and to resolve the problems. These may include:

• Programme management at level 1
• The Master Production Schedule meeting at level 2
• And perhaps a start of shift / sunrise team meeting at a lower level.

In the case of a food manufacturer with 8-hour customer required lead-times, a meeting was held between the production supervisor and the production planner every hour.

Conclusion

You may not be short of capacity. You may be short of a capacity planning process at any any of these four levels. You may also be short of a capacity control process which should ensure that these plans are met, and which does not cause conflicts by employing faulty measurement giving rise to local optimums. For example a local control system which encourages output may simply produce unwanted inventory.

Resources are defined by their Capability, Availability, Reliability, Agility, Timeliness) ("CARAT"). But these are not the only influences on output. A more comprehensive list of the drivers of capacity would include: Specification, Culture, Organisation, Processes / methods, Environment & finally Resources ("SCOPER").

In preference to applying sophistication, you could employ simplicity as a means of improving output & workflow. We recommend the Capacity 4S's (which correspond to the four levels) as the lowest overall cost method of managing capacity:

1. Segment the process to create (virtual) cells (A virtual cell is made up of a group of (not necessarily co-located) resources designed to deliver a product (family) ideally customer facing and encompassing the whole supply chain (Also see "Organisation Redesign" below)
2. S&OP (Sales & Operations Planning) to ensure plans are "do-able"
3. Sequence the work for most efficient bottleneck processing & good workflow
4. SMED the set-up / make ready between tasks to reduce non-productive time on the bottleneck

We cover the key aspects of items two & three & four in M23 Capacity Management, & item one, the formation of virtual cells, SCOPER / CARAT Analysis & the 21 wastes in S02 Business Process Reengineering, & S03 Vision of a "World Class" Organisation.

___________________________________________________________

The following further best practice articles were also mentioned in this paper:

The following public training courses and in-house workshops cover capacity management:

The following 8 training courses cover best practice capacity management:

Level 1: Business Planning

1.      S04 “Strategic Capacity Management” provides a guide to Business Planning using modern capacity strategy best practice

Level 2: Sales & Operations Planning

2.      M04 "Participative Master Production Scheduling" describes high level capacity planning processes of sales and operations planning, but focused on operations and master production scheduling rather than sales planning (this course is designed in conjunction with M05 below)

3.      M05 "Simple Capacity Planning and Control" describes how to design, implement and operate simple Capacity Planning & Control Systems (this course is designed in conjunction with M04 above)

4.      SSC08 “Participative Development, Sales & Operations Management” takes an holistic view of development planning, sales planning & operations planning processes, and applies the principles of capacity management to non-manufacturing & manufacturing businesses

Level 3: Workflow Management / Scheduling

5.      M09 “Manufacturing Resources Planning” describes the MRP2 approach to capacity management (M08 "MRP1" is a prerequisite for this course)

6.      M02 "Advanced Scheduling Systems" describes scheduling theory & APS systems in detail

Level 3/ 4: Simple Scheduling / Process Settings / Speeds / Set Up / Make Ready

7.      M11 “Simple Ways To Maximize Output & Workflow” describes detailed ways of maximizing throughput at an operational level

Guide to Capacity Management

8.      M23 "Capacity Management" provides a guide to capacity management for beginners, drawing on key aspects from all of the above capacity planning & control courses

Also you may be interested in our range of operations management training including OM02 Managing & Improving Individual Skills & Overall Skill Levels,

or the creation of simple, effective, processes in S02 Business Process Reengineering, or an executive overview of key aspects of world class organisations in S03 Vision of a "World Class" Organisation

To discuss your consulting or training needs with one of our independent consultants or trainers please Contact Us.

 Home Page

© SM Thacker & Associates (Consultancy and Training Specialists) Version 9 May 2009