Thanapun Prasertrungruang and B.H.W. Hadikusumo
IntroductionConstruction companies, especially highway contractors, rely heavily on mechanisation. Interruption of this mechanical supply not only incurs the direct costs of labour, replacement parts and consumables, but also the indirect costs of workforce, equipment downtime, contract delay, possible loss of client goodwill and ultimately, loss of profit (Edwards et al., 1998). Construction equipment is thus an important key factor for improving the contractor’s ability to perform their work more effectively and efficiently (Day and Benjamin, 1991). There are a number of factors affecting the productivity of construction equipment. Some factors are uncomplicated to identify and quantify, whereas others are problematic and difficult to predict. Downtime resulting from machine breakdown during operations is one of the most unanticipated factors that have a substantial impact on equipment productivity and organisational performance as a whole (Schaufelberger, 1999). Indeed, machine downtime is the most significant problem in equipment management faced by highway contractors (Prasertrungruang and Hadikusumo, 2007). Previous studies have addressed the issue regarding downtime in many aspects, for instance, downtime classification (Pathmanathan, 1980; Vorster and De La Garza, 1990), quantification (Vorster and De La Garza, 1990; Nepal, 2001), and prediction (Edwards et al., 2002), but little effort has been made to investigate the causes and consequences of downtime, particularly from a dynamic perspective (Nepal and Park, 2004). In fact, practices and policies for equipment management have some of the most dramatic effects on downtime (Elazouni and Basha, 1996). Variation in practices regarding the flows of machine-related factors (e.g. capital equipment, operators, mechanics, spare parts, and information) over time is thus a root cause of the dynamics of downtime (Nepal and Park, 2004). However, attempts to investigate the underlying interdependencies between these less tangible factors (e.g. equipment management practices) and downtime, which control the dynamic mechanisms of the system, have been rare (Edwards et al., 2002). Complex dynamic behaviour and the interaction between equipment management practices and downtime can be characterised by several key aspects, including cause-effect relationships, multiple feedback loops, nonlinear relationships, time-delayed responses, and involving both quantitative and qualitative data (Sterman, 2000). Managing construction equipment successfully with the aims of minimising downtime and maximising profit is therefore challenging (Edwards et al., 1998). This research is therefore intended to highlight the key dynamic features of downtime and its influential factors, using them as a framework in developing system dynamics (SD) simulation. Further, this simulation model generates several policy recommendations. The scope of this study focuses mainly on the equipment management practices and downtime of small to medium contractors in Thailand’s construction industry. Since machine weight is one of the major indicators of downtime and maintenance cost (Edwards et al., 2000a, b, 2002), only five types of large heavy equipment for highway construction were selected in this study (seeTable I). Note that weight interval for each equipment type is also assigned in order to allow for machine generalisation.
Equipment management practices and downtime in construction
The construction industry is exposed to a variety of risks. Equipment failure is one of the major risks frequently occurring during construction that consequently causes expensive downtime. However, downtime can be affected by other factors as well. Those factors are project-related factors, equipment-related factors, crew-level factors, site-related factors, and force majeure (Nepal and Park, 2004). As the consequential impact of downtime is huge, contractors need to build their competency in managing construction equipment throughout a machine’s lifecycle: acquisition, operations, maintenance, and disposal. Key elements of equipment management practices that contractors need to consider include, for instance, procurement decision approach (equipment acquisition stage), safety and training programs (equipment operational stage), scheduling preventive maintenance inspection and standby repair-maintenance facilities (equipment maintenance stage), and equipment economic life and replacement decisions (equipment disposal stage). In order to minimise the effects of downtime, the contractors have several alternative actions to consider, such as seeking substitute equipment, waiting until the repair is finished, accelerating work pace, modifying the work schedule, and transferring crews to other operations (Nepal and Park, 2004). The consequences of downtime can be categorised into two groups: downtime cost and downtime duration. Downtime duration can be classified into two types: scheduled and unscheduled downtime. Scheduled downtime is a time period when equipment is not available due to a routine task (e.g. periodic maintenance), whereas unscheduled downtime is a machine failure period caused by breakdown or equipment malfunction (Elazouni and Basha, 1996). Downtime cost consists of two elements: tangible and intangible costs. Tangible costs (e.g. costs of labour, material, and other resources needed to repair equipment) are easy to determine. However, intangible costs (e.g. production losses from labour and associated machines, extended overhead costs, liquidated damages, late completion charges) are rather difficult to quantify (Pathmanathan, 1980).
Data analysis
Data collected from each company (case) were examined using within-case and cross-case analysis approaches (Eisenhardt, 1989). Within-case analysis approach was performed first to allow the unique patterns of each case to emerge, and cross-case analysis approach was then used to uncover the similarities and differences among the cases. By employing cross-case analysis approach, several generic feedback loop structures, representing overall dynamic behaviours of cause-effect relationships with time-delayed effects of the system across all company cases, could be launched. In this study, the SD approach was adopted. SD is a way of analysing the behaviour of complex socioeconomic systems to show how organisation and policy influence behaviour over time (Sterman, 2000). Note that opinions and comments from the selected experts had been incorporated in every step of the study in order to validate the outputs (e.g. generic feedback structures, generic SD model, and policy recommendations). That is, the outputs could not be accepted as valid without an agreement from the selected experts. Based on the generic feedback structures constructed, the generic SD simulation model was then created. This step includes the identification of stock and flow diagrams. Stock represents accumulated quantities, whereas flow controls the changing rate of quantity going into or out of stock (Park, 2005). Powersimw software was utilised to construct the model.
To check the credibility of the generic SD model, data collected from each of the five company cases were input into the model in order to generate five different applied SD models. Each of the applied SD models represents the equipment management system of one particular contractor case. The generic SD model could not be accepted as valid unless all of the applied SD models were capable of generating time-series outputs of selected variables similar to those plotted using historical data (reference mode) from each of the company databases. Once the generic SD modelling process had been completed, it was validated (e.g. using sensitivity analysis) until the model was satisfactory. Last, policy analysis was used to recommend improvements to the equipment management system.
Generic feedback structuresFeedback structures are essential in SD as they are not only the foundation on which quantitative SD model is built, but also a valuable device in describing and understanding the dynamics of the system (Coyle, 1996). In this section, five generic feedback structures developed based on the interview data from all five small to medium contractor cases are presented as follows.
1. Machine acquisition feedback structure
2. Machine operational feedback structure
3. Machine maintenance feedback structure
4. Machine disposal feedback structure
5. Machine downtime feedback structure
Generic SD model structure
The aim of this section is to illustrate key elements of the generic SD model. The model can be divided into five subsystems. Each of the subsystems comprises various sectors, which were constructed as referenced to their corresponding feedback loops. As mentioned earlier, this study focuses on only five types of heavy equipment. Each type of equipment was modelled separately in the simulation and then connected together in order to capture its interdependent behaviour, which in fact induces the dynamics of downtime. The conceptual basis of the model structure was preliminarily derived from literature: infrastructure project management model (Nguyen and Ogunlana, 2005), design-build project management model (Chritamara et al., 2002), manufacturing organisational model (Keating et al., 1999), and downtime in equipment management (Nepal, 2001). Model descriptions for each of the subsystems are now presented.
Resources subsystem
The resources subsystem is made up of three sectors: equipment, operators, and mechanics. Operators and mechanics represent workforces that are categorised into two types: skilled and unskilled workers. The number of operators sought is dependent on total equipment invested, whereas the number of mechanics sought is controlled by total repair work orders and machine budget status. The equipment sector (e.g. backhoe) comprises three major stocks: “invested backhoes”, “invested backhoes on site”, and “invested backhoes under repair”. In the simulation, “invested backhoes” represent total backhoes currently owned by the contractor. Investment for additional backhoes is made if the number of backhoes sought is greater than total invested backhoes the contractor currently has. Backhoes can flow out of the company either to the job site or through disposal. Once breakdown occurs during operations, failed backhoes then flow into “invested backhoes under repair” and stay there until the repair is completed.
Quality subsystemQuality in equipment management was modelled and disaggregated into various sectors. The maintenance quality sector is modelled as the ratio between maintenance cost and repair cost, whereas crew’s skill is defined as the ratio between the number of skilled workers and total workers.
Several sectors were modelled as an arbitrary scale of 0 per cent to 100 per cent instead of a formula. These sectors include equipment quality upon acquisition, spare parts quality, experience, supervision, and management commitment in proactive maintenance.
For the preventive maintenance effort sector, it is assumed that working hours of mechanics are divided into two parts: repair hours and preventive maintenance hours. This means that, if the mechanics spend much time on repair, the effort devoted to preventive maintenance is diluted.
Lastly, the machine defect sector presents an accumulation of equipment (e.g. backhoe) defects during the simulation. Level of backhoe defects is used as an indicator for breakdown events. Defects continually build up as the equipment is utilised. Further, other factors (i.e. equipment quality upon acquisition, operator’s skill, collateral damage, and spare parts quality) also affect machine defect generation. In the simulation, breakdown occurs once defects reach 100 per cent. However, defects can be partly eliminated by performing two tasks: repair and preventive maintenance.
Performance subsystem
In the performance subsystem, a number of factors influence machine productivity, such as operator schedule pressure, fatigue, supervision, experience, machine defects, and machine reliability. Most of the effects of these factors on productivity are modelled on a qualitative arbitrary scale of 0 per cent to 100 per cent. For the machine availability sector, the formula of availability was given as the difference between total invested backhoes and invested backhoes under repair, divided by total invested backhoes. Alternately, for the machine reliability sector, the formula of reliability was defined as the discrepancy between invested backhoes on site and invested backhoes under repair, divided by total invested backhoes.
Machine efficiency is the last sector in this subsystem. Efficiency was modelled in this simulation to have an inverse relationship with machine defects. The more the machine defects, the less the machine efficiency will be.
Work pressure subsystem
This subsystem includes four sectors, namely, downtime cost pressure, operator schedule pressure, mechanics’ schedule pressure, and company workload. In the simulation, downtime cost pressure was defined as a percentage of total equipment cost. The higher the ratio between downtime cost and total equipment cost, the greater the downtime cost pressure. For the operator and mechanics’ schedule pressure sectors, the formulas are similar. Schedule pressure was defined in this study as the difference between workers (i.e. operators and mechanics) sought and the current number of workers, divided by total workers.
In the equipment workload sector of equipment (e.g. backhoe), workload is controlled by work creation rate and work completion rate. Work creation rate of backhoe is proportional to the difference between desired work scope capacity and current workload of backhoe. Alternately, work completion rate was defined as a multiplication of invested backhoes on site and expected backhoe productivity.
Financial subsystem
This subsystem comprises four sectors: equipment ownership cost, equipment operating cost, downtime cost, and machine budget status.
Equipment ownership cost is a fixed cost that is incurred each year whether the equipment is operated or not. This cost is made up of two elements: depreciation, and insurance and tax. Alternately, equipment operating cost is the cost incurred only when equipment is operated. Thus, operating costs vary with the amount of equipment used and job operating conditions. The major elements of operating cost include operator and labour wages, repair and maintenance costs, and fuel cost.
The downtime cost sector is a combination of equipment substitution cost, operator and labour idle cost, equipment idle cost, dependent equipment idle cost, dependent operator and labour idle cost, and repair cost of the failed machine.
Finally, for machine budget status, this sector is modelled as a stock. Machine budget status is increased if work progress is generated by equipment. Conversely, machine budget status is decreased as a result of higher downtime cost pressure as well as a greater ratio between machine investment and disposal rate.
Model testing for validation
After the data from each of the five contractor cases had been input into the generic SD model, five applied SD models were then adopted. Each of the applied SD models was subjected to a variety of validation tests to establish confidence in the soundness and usefulness of the generic SD model (Forrester and Senge, 1980). The study adopts model testing methodology from Sterman (2000), which has been categorised into two groups as follows.
1. Structural validation test
The model has been checked for the adequacy and appropriateness of its boundary throughout the modelling process by using data from many sources such as literature, interviews with experts (e.g. equipment managers), archival materials (e.g. machine investment and disposal records), and company databases (e.g. repair and maintenance costs). To clearly depict the boundary, model boundary chart and subsystem diagrams were also employed. Additionally, the feedback structures and the generic SD model derived during data collection were reassessed repeatedly with experts in the field to ensure they are consistent with reality.
2. Behaviour validation test
In order to assess the model’s ability to reproduce the behaviour of the real system, outputs from the simulation were compared with historical data obtained from the contractor. The comparisons of model base run and historical data (reference mode) of selected variables over time for a selected contractor case. It is obvious that behaviour of the base run and historical data is relatively similar. The model is thus successful in reproducing real data. In addition, a certain period of delay incident occurred after the policy intervention had also been found before the emergence of equipment performance improvement. Such delay incident was observed both in the model behaviour as well as in the real system. This confirms that the simulation model is behaviourally valid.
Policy formulation and analysis
Up to this point, the model has already been proven as structurally and behaviourally valid. This section thus aims at analysing and identifying a set of effective policies capable of improving equipment performance. Four key performance measures for comparing the policy behaviour were selected: total downtime duration, machine budget status, operator schedule pressure, and machine reliability. The policies were derived not only by reviewing literature such as (Laugen et al., 2005), but also from interviews with professionals. During the simulation, the following policies were experimented individually for a time frame of ten years (i.e. year 2005-2015). The policies can be categorised into four groups as shown in Table II.
Machine acquisition policies
Among the three equipment acquisition policies explored (Figure 1), machine quality upon acquisition (policy 3) is the best in term of producing lowest downtime duration and operator schedule pressure, as well as generating highest machine budget status and reliability. Machine fleet expansion (policy 2), in contrast, is the worst as it incurs highest downtime duration as well as lowest machine budget status and reliability. It is noted that policy 2 seems to generate lowest operator schedule pressure at the beginning, as more machines have been supplied. However, such positive behaviour of policy 2 lasts only for the first five years. In the long run, policy 2, in fact, tends to cause highest operator schedule pressure when works are interrupted due to more equipment failures.
IntroductionConstruction companies, especially highway contractors, rely heavily on mechanisation. Interruption of this mechanical supply not only incurs the direct costs of labour, replacement parts and consumables, but also the indirect costs of workforce, equipment downtime, contract delay, possible loss of client goodwill and ultimately, loss of profit (Edwards et al., 1998). Construction equipment is thus an important key factor for improving the contractor’s ability to perform their work more effectively and efficiently (Day and Benjamin, 1991). There are a number of factors affecting the productivity of construction equipment. Some factors are uncomplicated to identify and quantify, whereas others are problematic and difficult to predict. Downtime resulting from machine breakdown during operations is one of the most unanticipated factors that have a substantial impact on equipment productivity and organisational performance as a whole (Schaufelberger, 1999). Indeed, machine downtime is the most significant problem in equipment management faced by highway contractors (Prasertrungruang and Hadikusumo, 2007). Previous studies have addressed the issue regarding downtime in many aspects, for instance, downtime classification (Pathmanathan, 1980; Vorster and De La Garza, 1990), quantification (Vorster and De La Garza, 1990; Nepal, 2001), and prediction (Edwards et al., 2002), but little effort has been made to investigate the causes and consequences of downtime, particularly from a dynamic perspective (Nepal and Park, 2004). In fact, practices and policies for equipment management have some of the most dramatic effects on downtime (Elazouni and Basha, 1996). Variation in practices regarding the flows of machine-related factors (e.g. capital equipment, operators, mechanics, spare parts, and information) over time is thus a root cause of the dynamics of downtime (Nepal and Park, 2004). However, attempts to investigate the underlying interdependencies between these less tangible factors (e.g. equipment management practices) and downtime, which control the dynamic mechanisms of the system, have been rare (Edwards et al., 2002). Complex dynamic behaviour and the interaction between equipment management practices and downtime can be characterised by several key aspects, including cause-effect relationships, multiple feedback loops, nonlinear relationships, time-delayed responses, and involving both quantitative and qualitative data (Sterman, 2000). Managing construction equipment successfully with the aims of minimising downtime and maximising profit is therefore challenging (Edwards et al., 1998). This research is therefore intended to highlight the key dynamic features of downtime and its influential factors, using them as a framework in developing system dynamics (SD) simulation. Further, this simulation model generates several policy recommendations. The scope of this study focuses mainly on the equipment management practices and downtime of small to medium contractors in Thailand’s construction industry. Since machine weight is one of the major indicators of downtime and maintenance cost (Edwards et al., 2000a, b, 2002), only five types of large heavy equipment for highway construction were selected in this study (seeTable I). Note that weight interval for each equipment type is also assigned in order to allow for machine generalisation.
Equipment management practices and downtime in construction
The construction industry is exposed to a variety of risks. Equipment failure is one of the major risks frequently occurring during construction that consequently causes expensive downtime. However, downtime can be affected by other factors as well. Those factors are project-related factors, equipment-related factors, crew-level factors, site-related factors, and force majeure (Nepal and Park, 2004). As the consequential impact of downtime is huge, contractors need to build their competency in managing construction equipment throughout a machine’s lifecycle: acquisition, operations, maintenance, and disposal. Key elements of equipment management practices that contractors need to consider include, for instance, procurement decision approach (equipment acquisition stage), safety and training programs (equipment operational stage), scheduling preventive maintenance inspection and standby repair-maintenance facilities (equipment maintenance stage), and equipment economic life and replacement decisions (equipment disposal stage). In order to minimise the effects of downtime, the contractors have several alternative actions to consider, such as seeking substitute equipment, waiting until the repair is finished, accelerating work pace, modifying the work schedule, and transferring crews to other operations (Nepal and Park, 2004). The consequences of downtime can be categorised into two groups: downtime cost and downtime duration. Downtime duration can be classified into two types: scheduled and unscheduled downtime. Scheduled downtime is a time period when equipment is not available due to a routine task (e.g. periodic maintenance), whereas unscheduled downtime is a machine failure period caused by breakdown or equipment malfunction (Elazouni and Basha, 1996). Downtime cost consists of two elements: tangible and intangible costs. Tangible costs (e.g. costs of labour, material, and other resources needed to repair equipment) are easy to determine. However, intangible costs (e.g. production losses from labour and associated machines, extended overhead costs, liquidated damages, late completion charges) are rather difficult to quantify (Pathmanathan, 1980).
Data analysis
Data collected from each company (case) were examined using within-case and cross-case analysis approaches (Eisenhardt, 1989). Within-case analysis approach was performed first to allow the unique patterns of each case to emerge, and cross-case analysis approach was then used to uncover the similarities and differences among the cases. By employing cross-case analysis approach, several generic feedback loop structures, representing overall dynamic behaviours of cause-effect relationships with time-delayed effects of the system across all company cases, could be launched. In this study, the SD approach was adopted. SD is a way of analysing the behaviour of complex socioeconomic systems to show how organisation and policy influence behaviour over time (Sterman, 2000). Note that opinions and comments from the selected experts had been incorporated in every step of the study in order to validate the outputs (e.g. generic feedback structures, generic SD model, and policy recommendations). That is, the outputs could not be accepted as valid without an agreement from the selected experts. Based on the generic feedback structures constructed, the generic SD simulation model was then created. This step includes the identification of stock and flow diagrams. Stock represents accumulated quantities, whereas flow controls the changing rate of quantity going into or out of stock (Park, 2005). Powersimw software was utilised to construct the model.
To check the credibility of the generic SD model, data collected from each of the five company cases were input into the model in order to generate five different applied SD models. Each of the applied SD models represents the equipment management system of one particular contractor case. The generic SD model could not be accepted as valid unless all of the applied SD models were capable of generating time-series outputs of selected variables similar to those plotted using historical data (reference mode) from each of the company databases. Once the generic SD modelling process had been completed, it was validated (e.g. using sensitivity analysis) until the model was satisfactory. Last, policy analysis was used to recommend improvements to the equipment management system.
Generic feedback structuresFeedback structures are essential in SD as they are not only the foundation on which quantitative SD model is built, but also a valuable device in describing and understanding the dynamics of the system (Coyle, 1996). In this section, five generic feedback structures developed based on the interview data from all five small to medium contractor cases are presented as follows.
1. Machine acquisition feedback structure
2. Machine operational feedback structure
3. Machine maintenance feedback structure
4. Machine disposal feedback structure
5. Machine downtime feedback structure
Generic SD model structure
The aim of this section is to illustrate key elements of the generic SD model. The model can be divided into five subsystems. Each of the subsystems comprises various sectors, which were constructed as referenced to their corresponding feedback loops. As mentioned earlier, this study focuses on only five types of heavy equipment. Each type of equipment was modelled separately in the simulation and then connected together in order to capture its interdependent behaviour, which in fact induces the dynamics of downtime. The conceptual basis of the model structure was preliminarily derived from literature: infrastructure project management model (Nguyen and Ogunlana, 2005), design-build project management model (Chritamara et al., 2002), manufacturing organisational model (Keating et al., 1999), and downtime in equipment management (Nepal, 2001). Model descriptions for each of the subsystems are now presented.
Resources subsystem
The resources subsystem is made up of three sectors: equipment, operators, and mechanics. Operators and mechanics represent workforces that are categorised into two types: skilled and unskilled workers. The number of operators sought is dependent on total equipment invested, whereas the number of mechanics sought is controlled by total repair work orders and machine budget status. The equipment sector (e.g. backhoe) comprises three major stocks: “invested backhoes”, “invested backhoes on site”, and “invested backhoes under repair”. In the simulation, “invested backhoes” represent total backhoes currently owned by the contractor. Investment for additional backhoes is made if the number of backhoes sought is greater than total invested backhoes the contractor currently has. Backhoes can flow out of the company either to the job site or through disposal. Once breakdown occurs during operations, failed backhoes then flow into “invested backhoes under repair” and stay there until the repair is completed.
Quality subsystemQuality in equipment management was modelled and disaggregated into various sectors. The maintenance quality sector is modelled as the ratio between maintenance cost and repair cost, whereas crew’s skill is defined as the ratio between the number of skilled workers and total workers.
Several sectors were modelled as an arbitrary scale of 0 per cent to 100 per cent instead of a formula. These sectors include equipment quality upon acquisition, spare parts quality, experience, supervision, and management commitment in proactive maintenance.
For the preventive maintenance effort sector, it is assumed that working hours of mechanics are divided into two parts: repair hours and preventive maintenance hours. This means that, if the mechanics spend much time on repair, the effort devoted to preventive maintenance is diluted.
Lastly, the machine defect sector presents an accumulation of equipment (e.g. backhoe) defects during the simulation. Level of backhoe defects is used as an indicator for breakdown events. Defects continually build up as the equipment is utilised. Further, other factors (i.e. equipment quality upon acquisition, operator’s skill, collateral damage, and spare parts quality) also affect machine defect generation. In the simulation, breakdown occurs once defects reach 100 per cent. However, defects can be partly eliminated by performing two tasks: repair and preventive maintenance.
Performance subsystem
In the performance subsystem, a number of factors influence machine productivity, such as operator schedule pressure, fatigue, supervision, experience, machine defects, and machine reliability. Most of the effects of these factors on productivity are modelled on a qualitative arbitrary scale of 0 per cent to 100 per cent. For the machine availability sector, the formula of availability was given as the difference between total invested backhoes and invested backhoes under repair, divided by total invested backhoes. Alternately, for the machine reliability sector, the formula of reliability was defined as the discrepancy between invested backhoes on site and invested backhoes under repair, divided by total invested backhoes.
Machine efficiency is the last sector in this subsystem. Efficiency was modelled in this simulation to have an inverse relationship with machine defects. The more the machine defects, the less the machine efficiency will be.
Work pressure subsystem
This subsystem includes four sectors, namely, downtime cost pressure, operator schedule pressure, mechanics’ schedule pressure, and company workload. In the simulation, downtime cost pressure was defined as a percentage of total equipment cost. The higher the ratio between downtime cost and total equipment cost, the greater the downtime cost pressure. For the operator and mechanics’ schedule pressure sectors, the formulas are similar. Schedule pressure was defined in this study as the difference between workers (i.e. operators and mechanics) sought and the current number of workers, divided by total workers.
In the equipment workload sector of equipment (e.g. backhoe), workload is controlled by work creation rate and work completion rate. Work creation rate of backhoe is proportional to the difference between desired work scope capacity and current workload of backhoe. Alternately, work completion rate was defined as a multiplication of invested backhoes on site and expected backhoe productivity.
Financial subsystem
This subsystem comprises four sectors: equipment ownership cost, equipment operating cost, downtime cost, and machine budget status.
Equipment ownership cost is a fixed cost that is incurred each year whether the equipment is operated or not. This cost is made up of two elements: depreciation, and insurance and tax. Alternately, equipment operating cost is the cost incurred only when equipment is operated. Thus, operating costs vary with the amount of equipment used and job operating conditions. The major elements of operating cost include operator and labour wages, repair and maintenance costs, and fuel cost.
The downtime cost sector is a combination of equipment substitution cost, operator and labour idle cost, equipment idle cost, dependent equipment idle cost, dependent operator and labour idle cost, and repair cost of the failed machine.
Finally, for machine budget status, this sector is modelled as a stock. Machine budget status is increased if work progress is generated by equipment. Conversely, machine budget status is decreased as a result of higher downtime cost pressure as well as a greater ratio between machine investment and disposal rate.
Model testing for validation
After the data from each of the five contractor cases had been input into the generic SD model, five applied SD models were then adopted. Each of the applied SD models was subjected to a variety of validation tests to establish confidence in the soundness and usefulness of the generic SD model (Forrester and Senge, 1980). The study adopts model testing methodology from Sterman (2000), which has been categorised into two groups as follows.
1. Structural validation test
The model has been checked for the adequacy and appropriateness of its boundary throughout the modelling process by using data from many sources such as literature, interviews with experts (e.g. equipment managers), archival materials (e.g. machine investment and disposal records), and company databases (e.g. repair and maintenance costs). To clearly depict the boundary, model boundary chart and subsystem diagrams were also employed. Additionally, the feedback structures and the generic SD model derived during data collection were reassessed repeatedly with experts in the field to ensure they are consistent with reality.
2. Behaviour validation test
In order to assess the model’s ability to reproduce the behaviour of the real system, outputs from the simulation were compared with historical data obtained from the contractor. The comparisons of model base run and historical data (reference mode) of selected variables over time for a selected contractor case. It is obvious that behaviour of the base run and historical data is relatively similar. The model is thus successful in reproducing real data. In addition, a certain period of delay incident occurred after the policy intervention had also been found before the emergence of equipment performance improvement. Such delay incident was observed both in the model behaviour as well as in the real system. This confirms that the simulation model is behaviourally valid.
Policy formulation and analysis
Up to this point, the model has already been proven as structurally and behaviourally valid. This section thus aims at analysing and identifying a set of effective policies capable of improving equipment performance. Four key performance measures for comparing the policy behaviour were selected: total downtime duration, machine budget status, operator schedule pressure, and machine reliability. The policies were derived not only by reviewing literature such as (Laugen et al., 2005), but also from interviews with professionals. During the simulation, the following policies were experimented individually for a time frame of ten years (i.e. year 2005-2015). The policies can be categorised into four groups as shown in Table II.
Machine acquisition policies
Among the three equipment acquisition policies explored (Figure 1), machine quality upon acquisition (policy 3) is the best in term of producing lowest downtime duration and operator schedule pressure, as well as generating highest machine budget status and reliability. Machine fleet expansion (policy 2), in contrast, is the worst as it incurs highest downtime duration as well as lowest machine budget status and reliability. It is noted that policy 2 seems to generate lowest operator schedule pressure at the beginning, as more machines have been supplied. However, such positive behaviour of policy 2 lasts only for the first five years. In the long run, policy 2, in fact, tends to cause highest operator schedule pressure when works are interrupted due to more equipment failures.
Machine operational policies
For machine operational policies (Figure 2), in term of downtime reduction, multi-skilled training (policy 4) is the best, whereas work incentive scheme (policy 6) is the worst. However, by considering other measures, quality improvement team (policy 5) seems to be superior to others. This implies that policy 4 is effective only for the short term. In the long run, though, policy 5 is the most sustainable strategy for improving equipment performance. This could be due to the reason that, once a quality improvement team is established (policy 5), equipment defects are then continuously eliminated with the efforts of all parties involved. For policy 4, as operators are allowed to use multiple machines, downtime is then apparently reduced but defects rapidly accumulate, causing inferior performance in the long term.
For machine operational policies (Figure 2), in term of downtime reduction, multi-skilled training (policy 4) is the best, whereas work incentive scheme (policy 6) is the worst. However, by considering other measures, quality improvement team (policy 5) seems to be superior to others. This implies that policy 4 is effective only for the short term. In the long run, though, policy 5 is the most sustainable strategy for improving equipment performance. This could be due to the reason that, once a quality improvement team is established (policy 5), equipment defects are then continuously eliminated with the efforts of all parties involved. For policy 4, as operators are allowed to use multiple machines, downtime is then apparently reduced but defects rapidly accumulate, causing inferior performance in the long term.
Machine maintenance policies
As shown in Figure 3, it is obvious that shop crew capacity expansion (policy 7) is the worst, based on the four performance measures. Instead of expanding shop crew capacity, the contractor can use repair outsourcing strategy (policy 10), which is much more powerful, especially in minimising downtime duration and maximising machine reliability. Supplier strategy (policy 8) and autonomous maintenance (policy 9) can also be considered as effective policies since they produce a significant improvement superior to the base run, especially in minimising operator schedule pressure and maximising machine budget status. Although allowing equipment operators to perform simple maintenance tasks (policy 9) seems to increase their schedule pressure at first glance, the benefit of this policy, which is better machine condition, far outweighs this pitfall. With better equipment condition, operators can work productively without breakdown interruption, thus causing less schedule pressure to operators.
Machine disposal policiesFigure 4 shows that all three equipment disposal policies have superior performance to the base run. Dismantling-for-parts disposal (policy 13) seems to be the best, according to the four performance measures; while disposing of machines based on resale value (policy 11) produces the least benefits. Trade-in-for-new disposal (policy 12) is also a best candidate since it generates behaviours relatively close to those of policy 13. By dismantling some of the disposed machines for parts (policy 13), the contractor can save a significant amount of spare parts cost as well as shortening downtime duration since spare parts lead time becomes minimal.
Conclusions
This research is of value not only in facilitating more understanding regarding the dynamics of downtime for small to medium highway contractors, but also in assisting the contractors towards achieving reduced downtime and improved equipment performance by means of various policy recommendations using the SD approach.
.jpg)
To be successful in minimising downtime, equipment management practices must be viewed as an integration of multiple dynamic processes, which are all interrelated with downtime. In fact, downtime suffers heavily from the reinforcing cycles of operator and mechanics’ schedule pressure creep, as well as schedule disruption and downtime cost pressure growth. Although the balancing cycles of operator and mechanics’ skill improvement can alleviate downtime problem, their expected benefits always accrue after a delay, thus retarding the effect of improvement, or sometimes worsening the scenarios if contractors decide to discontinue training. Once the reinforcing cycles of downtime cost pressure and operator schedule pressure creep have been activated, proactive maintenance efforts are gradually diluted due to increasing downtime.
However, an increase in downtime can be controlled by adjusting the maintenance budget required for mechanics, facilities, parts, etc.
Having identified the dynamics of equipment management practices and downtime through five key feedback structures, a generic SD model was then constructed and successfully validated with five contractor cases. Results of the policy analysis reveal several promising policies, including machine quality upon acquisition (acquisition stage), quality improvement team (operational stage), repair outsourcing (maintenance stage), and dismantling for parts (disposal stage).
Future research could be directed towards exploring the dynamics of downtime associated with other factors in different perspective, rather than only with equipment management practices. Influences and impacts of such differences on the organisational performance are of prime interest. Further study could also focus on studying the dynamics of downtime in other types of contracting companies where equipment is a major resource used in production.
This research paper was published in the journal of “Engineering, Construction and Architectural Management, Vol. 15 No. 6, 2008, pp. 540-561”. Full paper is available upon request.
AbstractPurpose – Downtime resulting from equipment failure is a major problem consistently faced in highway construction. Since managing construction equipment is tightly connected to various activities and parties inside as well as outside of the firm, failure to account for this fact invariably causes downtime to be even more severe. Variation in equipment management practices is thus, indeed, a root cause of the dynamics of machine downtime. This study is intended to address key dynamic features of heavy equipment management practices and downtime in small to medium highway contracting firms and propose policies for equipment performance improvement.
Design/methodology/approach – Face-to-face interviews with equipment managers from five different small to medium highway construction companies in Thailand were conducted. Data were analysed using a system dynamics (SD) simulation approach.
Findings – To overcome downtime problems, contractors need to understand the dynamics of downtime as well as its influential factors, and thus manage their equipment as a dynamic process rather than one that is static. Based on the simulation, various policies are proposed to improve the performance of heavy equipment for small to medium highway contractors.
Originality/value – The research is of value in facilitating better understanding on the dynamics of equipment management practices and downtime as well as their interdependency.
Keywords Main roads, Production equipment, System analysis, Dynamics, Construction industry, Thailand
Paper type Research paper
.jpg)
.jpg)
