The Cobacabana system (Control of Balance by Card Based Navigation) with dynamic capacities

The Cobacabana system (Control of Balance by Card Based

Navigation) with dynamic capacities

Internal research for the master thesis of economics

___________________________________________________________________________

Author: Willem van der Klis

Student number: 1228250

Study program: Economics (Specialization Operations & Supply Chains) Faculty of Economics

University of Groningen

Cluster: Production management

Supervisors: Dr. M.J. Land

Dr. J. Riezebos

The Cobacabana system (Control of Balance by Card Based

Navigation) with dynamic capacities

Abstract

1. Introduction

Production companies have to cope with an increasingly turbulent environment. Not only large companies but also small- and medium-sized enterprises (SMEs) in the make-to-order (MTO) sector are faced with competitive challenges that foster the need for changes in product mix and volume, production rate changes, and shorter delivery periods (see Henrich et al. (2004)).

Logistics, planning and control of daily processes contribute to the fulfilling of company goals. Within the MTO companies the production planning and control (PPC) systems supports the production management. According to Stevenson, Hendry and Kingsman (2005), the choice of a PPC technique is a crucial strategic decision often based on superficial software features and insecure data.

To improve the PPC system the support of Enterprise Resource Planning (ERP) can be considered. However, the planning and control modules of ERP-systems can hardly be used within a job shop. These companies need simple planning and control systems to prevent them from turning back to legacy systems (Muda and Hendry (2002)). A possibility is the use of card-based systems such as Kanban and Polca (see Nicholas (1998) and Suri (1998)). Card-card-based systems are simple and effective PPC techniques but, according to Stevenson et al. (2005), the developed card-based systems do not meet the requirements of job shops. Improved card-based PPC techniques are needed to support the production management of small MTO companies.

A new card-based PPC technique called Cobacabana (Control of Balance by Card Based Navigation) has been developed for the production management of small MTO companies. The Cobacabana system is based on the Work Load Control (WLC) concept. The WLC concept is particularly suitable for SMEs in the MTO sector according to Hendry and Kingsman (1989) and research of Henrich, Land and Gaalman (2004) suggests that the WLC concept fits the typical characteristics of job shop manufacturing. The applicability of the WLC concept is determined by two distinguishing elements: ‘the buffering of the shop floor against fluctuations’ and ‘the use of aggregate norms’. By comparing aggregate workloads with a norm, the WLC concept is able to realize predictable throughput times based on the relation between workload and throughput times (Kingsman (2000)). Research by Oosterman et al. (2000) shows that focusing on the direct workload might create undesirable effects. Corrected aggregate workloads are a more appropriate variable to control and do not require adjustments. Land (2006) used the development of robust corrected aggregate workload norms to translate the WLC concept into a card-based system: Cobacabana.

preferences) or can fluctuate due to internal difficulties (for example absenteeism of personnel). Thus, a major issue for the Cobacabana system is to cope with dynamic capacities.

2. The Cobacabana system

Cobacabana is a card-based system that is designed for job shop companies to fulfill the PPC function. For this purpose, the system uses two card loops: the loop of order acceptance cards and the loop of order release cards. The loops of order acceptance cards are used to support the order acceptance and due date promising function. The loops of order release cards are used to authorize the release of new orders and aims at balancing the workload to provide predictable throughput times. The focus of this study is on the order release function of the Cobacabana system as capacity changes will have a direct effect on the predictability of the throughput times. In this section, the order release function of the Cobacabana system as introduced by Land (2006) will be discussed as well as some critical notes.

The Cobacabana system makes use of card loops to control the order release to the work floor. Figure 1 shows an example of the card loops for a job with routing: sawing, turning, drilling, and finishing.

Figure 1: Cobacabana release card loops between release and workstations.

The number of cards in the system relate to the workload and thereby strongly to the throughput time of the station (see Land (2004)). By fixing the maximum workload of a station to 100% (the norm), it is possible to apply a set of cards for this norm. One card represents a part (for instance 5%) of the total 100% that is available. An order is released when there are enough cards available to represent the contribution of the order to the workload of each station in the routing or, on estimate, the contribution of the order to the planned throughput time of each station in the routing.

Sawing Finishing Drilling Turning 20% 40% 60% 80% 100% Figure 2: The display with cards.

The display (figure 2) gives an overview of the work floor situation. Every card represents a percentage of the corrected aggregate workload. The empty card slots represent the released cards and show the percentage that is already used. The available cards, in figure 2 shown as cards with a pattern, represent the percentage that is available for release. After release, the workload of all the stations is balanced when the unreleased cards are evenly distributed near the norm value of 100%. Furthermore, potential bottlenecks can be easily identified with the display. For example, figure 2 shows that the sawing operation has still resources to spare and that the drilling operation can be restrictive for further order release.

The aim for balanced workloads stems from the WLC concept which is the root of the Cobacabana system. The philosophy of the WLC concept is to create predictable and short throughput times to improve the timing of order release, for quoting of realistic delivery times and for a good timing of capacity adjustments (Henrich (2005)). A predictable short throughput time at a workstation is enabled by keeping its direct load at a constant level. The direct load of a station is the sum of processing times of the jobs that are waiting or already being processed at the station.

The release of a job will not directly increase the direct load of a station, unless the station is the first in the routing of the job. However, a job does directly contribute to the aggregate load of a station. Aggregate loads incorporate all work upstream of the station that is being considered. As shown by Land (2004) the control of aggregate loads does not result in control of direct loads if the routing mix fluctuates. Therefore, an adjustment or standardization has to be made to control the direct loads. As presented by Oosterman et al. (2000), the processing times pjs of a job j at station

s

∈Ujs u D u D s T T * * is considered to be the jobs contribution to the calculated direct load. The fraction is the ratio between the planned throughput time TsD

*

of the workstation

s

∑

∈Ujs

u D u

T* ) with Ujs as the set of stations in the routing of job j upstream

times upstream of the considered station s and the throughput time of the station itself. This results in a calculated direct workload LstD

~

for station s at time t according to equation (1):

(1)

∑

I := Indicator that takes the value 1 on the interval specified for t, else 0. R

j

t := time of release of job j.

C js

t := time of completion of job j at station s.

Further standardization can be realized by depreciating the workload of station s with the maximum output OsD

*

. The maximum output is determined for the planned station throughput time T_s*D and specified in processing time units. The standardization results in equation (2) for the workload Lst:

(2) 1 () 100% ) , [ * * * ⋅ ⋅ ⋅ =

∑

Equation (2) can be used as a workload norm for the Cobacabana system by setting L_st to 100%. The workload norm relates to the percentage that is available as release cards on the display (figure 2).

The contribution C_js of a job can be used to calculate the number of cards that have to be released as specified in equation (3):

(3) 1 * 100% * * ⋅ ⋅ = ∑ ∈ js T T D s js p O C js U u D u D s

the workload will be smaller. The result is that the routing position of a station is of influence to the overview given by the display of figure 2.

In addition, the use of planned station throughput times within the calculation of the workload norm (equation 2) and the calculation of the job’s contribution (equation 3) suggests that after release the planned station throughput times will be reached. The station throughput time depends on the workload of a station and its maximum output. When after release the maximum allowed workload is reached, then the actual station throughput time will approximate the planned station throughput time. This is the case if all the available cards are released. In practice, the situation that all cards are released is uncommon and thus the actual station throughput times will not equal the planned station throughput times. Figure 2 gives an example of this situation as for some stations a percentage of release cards is still available after the release moment.

An adjustment to the value of the maximum output can be considered to reach the maximum allowed workload and therefore the planned station throughput times. The maximum output is used for the calculation of the workload norm (equation 2) and for the calculation of the job’s contribution (equation 3) and thus influences the released workload. According to Schönsleben (2004) the maximum output, there indicated as rated capacity, is calculated as the product of theoretical capacity, work centre efficiency, and capacity utilization. Theoretical capacity is the maximum capacity or output of a station without corrections. The work centre efficiency or efficiency rate indicates to what extent the production targets are met. The capacity utilization is the intensity of the use of a resource. There are two distinct factors in capacity utilization: availability and tactical under load. The availability of a capacity takes in consideration all the possible downtimes due to breaks, cleaning tasks, breakdowns, unplanned absences, etc. Tactical under load means that the desired capacity utilization should be less than 100% to avoid long queue times by correction of the workload of the station. For the Cobacabana system it is needed to adjust the tactical under load factor from the calculation of the maximum output so that the maximum allowed workload can be reached and thus the planned station throughput times.

3. Self regulating capability of the Cobacabana system

The flow of cards must guarantee the release of new jobs and consequently a stable workload for the stations. As shown in figure 1, the flow of cards can be divided into two sub flows. The first sub flow consists of the cards that are attached to the order guidance form of a job and are en route to a specific station. The second flow consists of the cards returning to the Cobacabana display that have been detached from the order guidance form of the job at the specific station. The velocity of the flow of cards is determined by the cycle time. The cycle time is the period between the time of release of a card and the time that the card returns to the display. When a capacity increase occurs the cycle time will decrease and when a capacity decrease occurs the cycle time will increase. The fluctuations in the cycle time will result in changes in the velocity of the flow of cards and consequently in the amount of jobs that are released. A capacity increase results in speeding up the release of jobs and a capacity decrease slows down the release of jobs. Thus, the Cobacabana system has a self regulating capability to absorb capacity fluctuations by automatically adjusting the release of jobs and consequently the workload of the stations.

The self regulating capability of the system depends on the following three characteristics: (1) the extent of the capacity fluctuations, (2) the routing position of the station with fluctuating capacity, and (3) the position of bottleneck stations. The first characteristic, the extent of the capacity fluctuations, is defined by the duration and magnitude of the capacity increase or decrease. The magnitude is the change of capacity in production time units for the duration of the capacity fluctuation. Figure 3 shows a classification of the extent of the capacity fluctuations into minimal, moderate, and extreme. The boundaries are loose and represented by dotted lines because the specific situation of a company determines where the boundaries have to be placed. Short Long Duration Minimal Moderate Extreme

The extent of the capacity fluctuation will be classified ‘minimal’ when the effect on the station throughput time can be neglected with respect to fulfilling the promised due dates. If classified as ‘moderate’ the effect on the station throughput time is such that it can not be neglected. The classification ‘extreme’ is used for a capacity fluctuation that not only disturbs the throughput time of the station but also disturbs the throughput times of downstream stations such that the disturbances can not be neglected.

Not only is the extent of the capacity fluctuation of the station important to the self regulating capability but also the second characteristic: the routing position of the station. The Cobacabana system is designed for job shops which means that the routing position of a station can differ for each job. Within a pure job shop the jobs flow with undirected routings which means that the average routing position of all the stations is the same. It can also be that within the job shop a dominant flow is present. Then the jobs flow with directed routings which means that the average routing position of the station is relatively more upstream or relatively more downstream (Land (2004)). The capacity fluctuations will have less impact if the station is far downstream as it would have when the station is far upstream. In the latter case, more stations succeed the station with the capacity fluctuation and therefore more stations will be influenced.

The third characteristic is the position of bottleneck stations. If the station with the capacity fluctuation is succeeded by non-bottleneck stations then the capacity fluctuations can be absorbed with the spare capacity of the non-bottleneck stations. This is not the case for bottleneck stations. A capacity decrease at a preceding station of the bottleneck leads to less workload and can influence the bottleneck’s throughput time. A capacity increase at a succeeding station of the bottleneck leads to a decreased workload of the succeeding station because the capacity of the bottleneck station is insufficient to replenish the workload and therefore will influence the throughput time.

A final remark to the self regulating capability stems from the foundation of the Cobacabana system: the Workload Control concept. Cobacabana is based on long term averages for the determination of the parameters maximum output, planned throughput time, and maximum workload. The self regulating capability is able to automatically adjust the release of jobs based on those parameters. Only significant capacity changes should lead to adjustments of the parameters and not the small changes that can be regarded as noise. Furthermore, the classification of the extent of a capacity fluctuation (see figure 3) and especially the division of the duration into ‘short’ and ‘long’ is relative to the parameters of the Cobacabana system.

4. Adjustments for dynamic capacities

The order release function of the Cobacabana system is designed to balance the workload of the stations in order to achieve predictable throughput times. The workload together with the output determine the station throughput time. When capacity changes are introduced the output will change but not the workload. The result is that the actual station throughput time will deviate from the planned station throughput time. Therefore, the maximum workload norm has to be adjusted to deal with dynamic capacities.

Dynamic capacities are the changes to the production system due to capacity increases or decreases. Capacity increases are required if, focusing on the order release moment, the necessity occurs to release jobs that do not fit within the norms. Examples of capacity increases are the hiring of temporary workers, the use of extra machines, overtime, etc. Capacity decreases can occur unexpected and expected. Unexpected capacity decreases (such as machine breakdowns, illness, etc.) are not planned while expected capacity decreases are planned measures. The Cobacabana system is hedged against unexpected capacity decreases by the use of the capacity availability factor within the calculation of the maximum output and the increased cycle time of the loops of release cards. Within the calculation of the maximum output, as mentioned in section 2, the capacity availability factor takes in consideration the possible downtimes due to breaks, cleaning tasks, breakdowns, unplanned absences, etc. The Cobacabana system uses the loops of the release cards to control the order release. If an unexpected capacity decrease occurs, the cycle time between the release of the card and return of the card to the planning display will increase. The increased cycle time automatically slows down the release of jobs to the work floor which is desirable in case of a capacity decrease (see section 3).

4.1. Simple case: output rate change for one job

If the necessity occurs to release a job that does not fit within the norms, a company can decide to increase the capacity. A simple solution would be to allow the release of a job that actually has to be rejected based on the norm of 100% and to create an additional release card for the percentage in excess of 100%. The additional release card is order specific and can be used to display the percentage but also to specify the commitment and details of the needed capacity increase. The additional release card returns to the planner when the job is finished and has to be discarded after first use. In this paper, a job that has to be rejected based on the norm values will be called a ‘critical job’. Unfortunately, the problem is that by increasing the workload in excess of 100% the capacity will be insufficient to maintain planned throughput times. The solution is to increase the capacity of a station to make sure that a critical job can be released. The capacity can be changed by, for example, the use of overtime, the hiring of temporary workers or by outsourcing. The capacity increase will result in an increase of the output of the station and therefore more jobs can be released.

The maximum output is determined for the planned station throughput time TsD

*

and specified in processing time units. To determine when, for which period and to what level the capacity has to be increased it is convenient to transform the output into an output rate that is time dependent. The planned output rate RsD

*

of a station is the division of the maximum output and the planned station throughput time:

(4) _D s D s D s T O R _* * * =

The output rate can be increased by increasing the capacity for a period of time. Equation (5) gives the needed output rate R_jst of a station at time t after the critical job j is released. fjsis the part of pjs that increases the workload above 100% and thus the additional capacity that is needed. The duration of the capacity increase is the interval between the start time (t1_js) and end time (t2_js) of the capacity increase.

(5) _{* 2 1 [ , )} * 2 1 ) ( js jst t js js js D s D s jst I t t t f T O R ⋅ − + = With: ) , [1 2 ) ( js jst t t

The interval of the capacity increase can be of influence to the throughput time of the critical job and other jobs. Only if the interval of the capacity increase starts and ends during the period that the critical job is processed at the station the throughput times will be as planned, because the additional capacity is directly used for the critical job and not for other jobs.

4.2. Output rate change for the general case

It is unlikely that the capacity increase of a station can be phased with the processing period of one critical job. First, it is most likely that more than one critical job will be released. To phase the capacity increases with a set of critical jobs will be difficult. Second, the capacity increase is constrained (available personnel, capacity limitations at stations, etc.) which can be a problem with respect to determination of the duration and moment in time. The same restrictions also count for the expected capacity decreases. It will be more convenient to adjust the order release function for a set of critical jobs instead of altering the interval of the capacity change. The set of critical jobs is the set with one ore more jobs that have to be rejected based on the norm values but can be released due to the capacity change.

For the adjustment of the order release function it is convenient to use the output rate that results from the capacity changes. The output rate Rst of station

s

t

n which starts at time t1ns and ends at time

2

ns

t . The capacity changes can be simultaneous and therefore have to be summed.

(6) _{* 2 1 [ , )} * 2 1 ) ( ns nst t N n ns ns ns D s D s st I t t t f T O R _⋅ − + =

∑

I := Indicator that takes the value 1 on the interval specified for t, else 0.

4.3. Timing issues for the order release relative to the output rate change

both curves estimates the otherwise discrete course of the curves and is therefore an estimated average for the input and output.

1 s t t_s2 C u m u la tiv e p ro c e s s in g tim e u n it s D s T* D st L 1 s T 1 st L 1 R t Q s T* 1 st L LD_st 1 s T T_s*D

Figure 4: Throughput diagram of scenario 1 for a station with a capacity increase.

1 s t t_s2 Time (t) C u m u la ti v e p ro c e s s in g tim e u n it s D s T* D st L 2 s T 2 st L 2 R t Input Output D s Q s T T* + * Scenario 2: 2 st L equals D st L 2 s T larger than D s T* 2 S t

For scenario 1 (figure 4) are the critical jobs released in such a way that the first critical job arrives at the station’s queue at the moment that the capacity increase starts. This is done by releasing the critical jobs at moment tR1 such that the time period until the start of the capacity increase t1_s is equal to the expected average throughput time up to the station’s queue T_s*Q. The release of critical jobs ends at the moment such that the slope of the input curve equals the slope of the output curve at ts2. This is done by ending the release of critical jobs at

Q s

s T

t2₋ * . Figure 4 shows that the direct load L1_st on the interval of the capacity increase is the same as the direct load LDst for the interval without a capacity increase. However, the throughput time decreases from T_s*D toT_s1.

The second scenario, represented in figure 5, is able to keep the throughput time stable by increasing the direct load of the station. The critical jobs are released at moment tR2 such that they arrive at the station’s queue at time tS2 with a period of one planned station throughput time before the capacity increase starts (tR2 t_s1 T_s*Q T_s*D

− −

= ). The release of critical job ends

with a period sD

Q

s T

T* ₊ * before the end time of the capacity increase ts2. The result of the second scenario is that the direct load of the station is increased from LD_stto L2_st but that the planned station throughput time TsD

*

is equal to the throughput time during the interval of the capacity increase T_s2. Thus, the throughput diagram of figure 5 shows that to keep the throughput times stable it is needed to release the critical jobs in such a way that the first critical job arrives at the queue of the station with a period equal to the planned station throughput time before the start time of the capacity increase.

The same adjustment of the order release function as described by scenario 2 works for expected capacity decreases (see figure 6). The dotted curve represents the input and the continuous curve the output. The capacity decrease starts at time t1s and ends at time

2

s t . Throughput times can be held stable by releasing fewer jobs such that the decrease of the slope of the input curve starts at time tS3 one planned station throughput before the capacity decrease starts. This is achieved by releasing fewer jobs at release moment tR3 with a period of

D s Q

s T

T* + * before the start of the capacity decrease (tR3 =t_s1−T_s*Q−T_s*D). The release of fewer jobs ends with a period T_s*Q T_s*D

1 s t ts2 Time (t) C u m u la ti ve p ro ce ssi n g ti m e u n it s D s T* D st L 3 s T 3 st L 3 R t Input Output 3 S t Scenario 3: 3 st L LD_st 3 s T smaller than D s T* equals

Figure 6: The throughput diagram for a station with an expected capacity decrease.

From the throughput diagrams it can be concluded that for capacity increases and expected capacity decreases it is needed to foresee the changes in advance with at least a time period greater than T_s*Q+T_s*D. Unfortunately, the information on the Cobacabana-display (figure 2) is insufficient for this purpose. The display shows the current work floor situation at order release but not a projection of capacity changes over time. A capacity profile can be used to obtain the needed information at the moment of order release.

4.4. A capacity profile and planning fence to improve the order release

The capacity profile of figure 7 represents the capacity of a station over time (see Schönsleben (2004)). The profile is plotted with a rectangular curve to simplify the figure. A planning fence is added to depict the future time period of the capacity profile that is included in the order release decision. For release moment 0 the corresponding planning fence is F_0s and for release moment 1 the corresponding planning fence is F1_s. The number of release cards has

Order release moment (m) 1 2 3 4 5 6 7 8 9 10 11 25% 50% 75% 175% 150% 125% 100% s F1 s F0 0

Figure 7: The capacity profile of a station.

As noted, the input of a station is phased with a capacity increase or expected capacity decrease when the order release of the station is altered with a time period of the sum of the average expected up to station’s queue throughput time and planned station throughput time before the capacity change starts. Therefore, the start t_msS of the planning fence for station s at the order release moment with index m has to be the time of the order release tmsR summed with the average up to station’s queue throughput time T_s*Q and one planned station throughput time TsD

*

:

(7) t_msS =t_msR +T_s*Q+T_s*D

To avoid double count of a capacity change it is necessary to end the planning fence before the start time of the next planning fence. Equation (8) depicts the end time t_msE of the planning fence of order release moment m:

(8) sD Q s R s m S s m E ms t t T T t = ( +1) = ( +1) + * + *

counts for the set of unreleased jobs. Statistical analysis of the up-to-station’s queue throughput times can be useful to determine the start of the planning fence for a set of jobs.

The start time t_msS and end time t_msE of the planning fence F_ms define the interval T_ms as [t_msS ,t_msE ). Only the capacity changes within the interval have to be used for the order release at the release moment. Looking back to figure 7 shows that only the light shaded area equivalent to a capacity increase is within the interval of the current release moment 1 and can be used for the release of additional jobs. The dark shaded area equivalent to a capacity decrease is already offset at the previous release moment.

4.5. Adjustment of the workload norm

It can be that a planned capacity change is before the start time of the current planning fence but not offset at an earlier release moment. The capacity change will influence the station throughput times but can not be offset by adjusting the current order release decision. Section 3 proved that the Cobacabana system has a self regulating capability to absorb a capacity change. If the self regulating capability is insufficient other options can be considered like capacity increases to offset capacity decreases or the release of jobs with short up to station’s queue throughput times. The planner has to judge if and what kind of measures are necessary. For the current release decision only the planned capacity changes that are within the planning fence and which are not a measure to other capacity changes are of importance.

The variable ams is defined as the average capacity change in processing time units for station s at release moment m that is within the interval of the planning fence. At the release moment m is the variable a_ms the change in the allowed quantity of jobs to be released in terms of processing time units. For the calculation of the workload Lst (see equation 2) are the processing times pjs for every job j corrected with the factor ∑

∈ ⋅ js U u D u D s T T D s O * * * 1 . The up-to-station throughput time

∑

T* includes the planned station throughput time of station s and is

defined for a specific job j (see section 2). The summation of the up-to-station queue time T_s*Q and the planned station throughput time TsD

*

is an average value for

∑

∈Ujs u D u T* . The variable ams

can be related with the workload L_st by correcting a_ms with _D s Q s D s D s T T T O * * * * 1 + ⋅ . This results in a

(9) 1 _{* *} 100% * * ⋅      + ⋅ + = D s Q s D s D s ms ms T T T O a A

The percentage Ams has to be used as the new station’s norm on the Cobacabana-display instead of the norm value of 100%. If the percentage is less than 100%, then the percentage available as release cards decreases. If the percentage is more than 100% the percentage available as release cards increases. The percentage change of release cards relates to the change in the allowed quantity of jobs to be released in terms of processing time units. The percentage above 100% becomes available by creating additional release cards. The additional release cards can be used as regular release cards but have to be discarded after first use. The problem can occur that the flow of regular release cards is disturbed by the additional release cards. A piling-up of additional release cards will temporarily freeze the flow of regular release cards. It is best to mix the additional release cards with the regular release cards to minimize the disturbance.

5. Long-term adjustments

The Cobacabana system has to be adjusted when capacity increases or decreases occur that are permanent or for a long-term. There are two solutions: (1) to make the change of additional cards or withdrawn cards, as introduced in section 4, permanent or (2) to recalculate and adjust the Cobacabana system to the new situation.

To make the change of additional cards or withdrawn cards permanent means that the changes are incorporated in the Cobacabana system. The additional cards or withdrawn cards become part of the Cobacabana display. The advantage is that the effort needed to change the system is minimal. The disadvantage is that the operational characteristics (the maximum output

D s O*

, the job’s contribution Cjs, and the workload Lst) are changed but not within the Cobacabana system. A norm of 100% does not correspond with the maximum workload after the permanent adjustment but to the percentage Ams as calculated with equation (9). This can lead to erroneous interpretation and use of the Cobacabana display.

The second solution is to redesign the Cobacabana system based on the new operational characteristics. The advantage is that the Cobacabana display is in line with the new operational characteristics and that the 100% norm corresponds to the maximum workload. The disadvantage is that the contributions (see equation 3) of all the jobs that make use of the changed station’s capacity have to be recalculated. This can be a time consuming task especially if capacity changes for other station’s capacities occur frequently.

6. Conclusion

In this study, the order release function of the Cobacabana system and its relation with dynamic capacities has been investigated. The study indicates that the Cobacabana system is able to absorb small capacity fluctuations by the self regulating capability of the system. The self regulating capability is able to speed up the release of jobs when a capacity increase occurs and to slow down the release of jobs when a capacity decrease occurs. However, the self regulating capability is unable to timely phase the order release with the capacity changes.

Capacity profiles with fixed planning fences have been introduced to aid the planner in adjusting the order release function. The capacity profile is used to foresee the capacity changes while the planning fence enables the planner to adjust the timing of the order release. The timing of the order release makes sure that the workload is adjusted in time and is phased with the capacity change. The adjustment of the workload results from the calculation of a new percentage for the workload norm. The new percentage is used on the Cobacabana display so that the amount of released jobs is sufficient to offset the capacity change. The result of the improved order release function is that the effects of capacity changes on the throughput times are minimized.

The improved order release function can also be used to cope with long-term or permanent capacity changes. Although, to redesign the Cobacabana system based on the new situation can be worthwhile to avoid erroneous interpretation and use of the Cobacabana display. The disadvantage is the substantial effort needed to redesign the system. For frequent capacity changes the improved order release function proves to be a simple and effective tool.

References

Hendry, L.C., Kingsman, B.G. 1989, Production planning systems and their applicability to make-to-order companies, in: European Journal of Operational Research, Vol. 40, pp. 1-15.

Henrich, P., Land, M.J., Gaalman, G.J.C. 2004, Exploring applicability of the workload control concept, in: International journal of production economics, vol. 90, pp. 187-198.

Henrich, P. 2005, Applicability aspects of workload control in job shop production, PhD Thesis, University of Groningen, Labyrint Publications, Ridderkerk, The Netherlands.

Kingsman, B.G. 2000, Modelling input-output workload control for dynamic capacity planning in production planning systems, in: International Journal of Production Economics, vol. 68, pp. 73-93.

Land, M.J. 2004, Workload control in job shops, grasping the tap, PhD Thesis, University of Groningen, Labyrint Publications, Ridderkerk, The Netherlands.

Land, M.J. 2006, Cobacabana : A card based system for job shop control, in: Pre-prints of the fourteenth International Working Seminar on Production Economics, vol. 3, pp. 165-174. (Submitted to the International Journal of Production Economics).

Muda, S., Hendry, L. 2002, Proposing a world-class manufacturing concept for the make-to-order sector, in: International Journal of Production Research, vol. 40, pp. 353-373.

Nicholas, J.M. 1998, Competitive Manufacturing Management: continuous improvement, lean production, customer-focused quality, Irwin/McGraw-Hill

Oosterman, B., Land, M.J., Gaalman, G.J.C. 2000, The influence of shop characteristics on workload control, in: International journal of production economics, Vol. 68, pp. 107-119.

Schönsleben, P. 2004, Integral Logistics Management: Planning & Control of Comprehensive Supply Chains, second edition, St. Lucie Press, Boca Raton, Florida

Suri, R. 1998, Quick Response Manufacturing: A Companywide Approach to Reducing Lead Times, Productivity Press, Portland, Oregon.