8
Decentralised site management— a case study Patrick Barber, Cyril Tomkins* and Andrew Graves Agile Construction Project, School of Management, University of Bath, Bath BA2 7AY, UK Where projects are complex, there may be a need for greater decentralisation of control in order to avoid the bureaucratic ineciencies of central direction. Decentralisation does not, however, dispense with the need to co-ordinate activities. This paper shows how a major UK construction company successfully implemented a balance between cellular autonomy and overall project risk management in a major road project. # 1999 Elsevier Science Ltd and IPMA. All rights reserved Keywords: project management, self organising project structures, project co-ordination, project risk management Introduction By granting more autonomy to segments of a complex operation, one gains a better utilisation of local knowl- edge, quicker responses to needs and improved motiv- ation to succeed through clearly identified responsibility (Anthony, et al. 1 ). On the other hand, decentralisation runs the risk of inconsistency and con- tradictory action unless activities can be co-ordinated eectively. There seem to be two main aspects which need to be considered in addressing how co-ordination may be maintained in decentralised organisations. First, Peters and Waterman 2 resolved the apparent paradox of exer- cising both tight and loose control simultaneously by stressing: ‘‘. . . autonomy is the product of discipline. The disci- pline (a few shared values) provides the framework. It gives people confidence (to experiment, for instance) stemming from stable expectations about what really counts.’’ (Peters and Waterman, 2 p. 322.) This management philosophy was clearly recognised by Jolivet and Navarre 3 who described decentralised management developments in Spie-Batignolles. They explained how the classic management approach invol- ving pyramidal and functional specification of tasks, standardisation of procedures and regulation by hier- archy gave way to autonomous teams built upon self organising principles and performance control through the use of a few meta-rules which set the values by which subordinate decisions had to be made. The second aspect of co-ordination required is the management of the risk profile for the whole organis- ation and how resource allocation is modified dynami- cally to allow for uncertainties and newly perceived risks as they arise. The question is how can this appar- ent need for centralisation be operationalised when maximum autonomy is sought for cellular forms of op- eration. This paper shows, by means of a specific case study, how a particular form of organisation, a major con- struction project, can be organised in a way which allows the use of decentralised multi-disciplinary teams without losing co-ordination of the project and, in par- ticular, how risk and adaptive resource allocation can be managed dynamically in such a setting. The case study The case investigated the decentralisation of control of the day to day work of a major road (A13) construc- tion project in the UK. The A13 was selected for study because it was known that: 1. it had adopted a multi-functional team based (cellu- lar) project structure, 2. these teams had been given considerable autonomy, 3. the project manager had part-way through the pro- ject introduced: . ‘‘a no blame culture’’ to encourage sub-ordinates to exercise fully their new independence and more truthful reporting, . a new form of project risk management. The A13 road improvement project entailed the con- struction of a completely new section of road involving 5.2 km of dual carriageway passing through land designated as a Site of Special Scientific Interest. The project included three major viaducts plus two major road interchanges, and had specific diculties as it also had to be built over a disused foundry and the slurry lagoons. These factors, plus the fact that the International Journal of Project Management Vol. 17, No. 2, pp. 113–120, 1999 # 1999 Elsevier Science Ltd and IPMA. All rights reserved Printed in Great Britain 0263-7863/99 $ - see front matter PII: S0263-7863(98)00014-3 *Contact person for correspondence. 113

Decentralised site management—a case study

Embed Size (px)

Citation preview

Page 1: Decentralised site management—a case study

Decentralised site managementÐa case study

Patrick Barber, Cyril Tomkins* and Andrew GravesAgile Construction Project, School of Management, University of Bath, Bath BA2 7AY, UK

Where projects are complex, there may be a need for greater decentralisation of control in orderto avoid the bureaucratic ine�ciencies of central direction. Decentralisation does not, however,dispense with the need to co-ordinate activities. This paper shows how a major UK constructioncompany successfully implemented a balance between cellular autonomy and overall project riskmanagement in a major road project. # 1999 Elsevier Science Ltd and IPMA. All rightsreserved

Keywords: project management, self organising project structures, project co-ordination, project risk management

Introduction

By granting more autonomy to segments of a complexoperation, one gains a better utilisation of local knowl-edge, quicker responses to needs and improved motiv-ation to succeed through clearly identi®edresponsibility (Anthony, et al.1). On the other hand,decentralisation runs the risk of inconsistency and con-tradictory action unless activities can be co-ordinatede�ectively.There seem to be two main aspects which need to be

considered in addressing how co-ordination may bemaintained in decentralised organisations. First, Petersand Waterman2 resolved the apparent paradox of exer-cising both tight and loose control simultaneously bystressing:

``. . . autonomy is the product of discipline. The disci-

pline (a few shared values) provides the framework. Itgives people con®dence (to experiment, for instance)stemming from stable expectations about what really

counts.'' (Peters and Waterman,2 p. 322.)

This management philosophy was clearly recognisedby Jolivet and Navarre3 who described decentralisedmanagement developments in Spie-Batignolles. Theyexplained how the classic management approach invol-ving pyramidal and functional speci®cation of tasks,standardisation of procedures and regulation by hier-archy gave way to autonomous teams built upon selforganising principles and performance control throughthe use of a few meta-rules which set the values bywhich subordinate decisions had to be made.The second aspect of co-ordination required is the

management of the risk pro®le for the whole organis-ation and how resource allocation is modi®ed dynami-

cally to allow for uncertainties and newly perceivedrisks as they arise. The question is how can this appar-ent need for centralisation be operationalised whenmaximum autonomy is sought for cellular forms of op-eration.This paper shows, by means of a speci®c case study,

how a particular form of organisation, a major con-struction project, can be organised in a way whichallows the use of decentralised multi-disciplinary teamswithout losing co-ordination of the project and, in par-ticular, how risk and adaptive resource allocation canbe managed dynamically in such a setting.

The case study

The case investigated the decentralisation of control ofthe day to day work of a major road (A13) construc-tion project in the UK. The A13 was selected for studybecause it was known that:

1. it had adopted a multi-functional team based (cellu-lar) project structure,

2. these teams had been given considerable autonomy,3. the project manager had part-way through the pro-

ject introduced:

. ``a no blame culture'' to encourage sub-ordinatesto exercise fully their new independence and moretruthful reporting,

. a new form of project risk management.

The A13 road improvement project entailed the con-struction of a completely new section of road involving5.2 km of dual carriageway passing through landdesignated as a Site of Special Scienti®c Interest. Theproject included three major viaducts plus two majorroad interchanges, and had speci®c di�culties as italso had to be built over a disused foundry and theslurry lagoons. These factors, plus the fact that the

International Journal of Project Management Vol. 17, No. 2, pp. 113±120, 1999# 1999 Elsevier Science Ltd and IPMA. All rights reserved

Printed in Great Britain0263-7863/99 $ - see front matter

PII: S0263-7863(98)00014-3

*Contact person for correspondence.

113

Page 2: Decentralised site management—a case study

area was inherently low lying and marshy, meant thatthe contract was always going to be a di�cult one tocomplete and exhibited just the situation in which theJolivet and Navarre matrix3 suggests that managementis best achieved through self organising cells based onmeta-rules. It was, therefore, not surprising that a del-egated approach should have been speci®ed as theapproach to adopt.

Revising the project management structure

When the site was set up, the project manager wasaware of an internal company report which identi®eddi�culties that some large projects had experiencedwith internal communications and planning.The classic or traditional form of managerial control

is from the top down with the tendency to operate inan authoritarian and adversarial manner. The work isbroken down into functional groups (e.g. road laying,structure, earth works and drainage and sub-functionswithin those groups) which attempt to produce thebest performance possible whilst keeping to their bud-gets. Performance is measured by means of local func-tional measures. These would be cubic metres of earthmoved, cubic metres of concrete laid, tons of steel®xed, etc. These measures show how much is beingmoved, laid or ®xed, but they do not show if it isbeing done in the right place or at the right time. Thefocus is upon e�ciency rather than overall goal e�ec-tivenessÐthe latter being left to central managementto worry about.Under such an approach, the internal report

suggested that di�erent functions had, on other pro-jects, pulled in di�erent directions attempting to meettheir performance targets as measured against theire�ciency performance measures. This often meant thatthe various functions got in each others' way and dis-putes would break out between the engineers in charge

of the various functions as their bonuses were deter-mined by these measures. With these problems inmind, it was decided to re-organise the project with a¯atter management structure based around auton-omous self managing teams which would be respon-sible for all operations in given geographical locationsof the project.The contrast between this management structure

and the traditional functional based method of organ-ising is shown in Figure 1: the previous vertical func-tional team orientation was replaced by the horizontalorganisation by areas. The performance of these multi-functional teams was still measured against measuressuch as cubic metres of earth moved or tons of steel®xed, as the company insisted on having them, butthey were used in a more intelligent way, designedspeci®cally to re¯ect each team's key operations andtheir relationship to the overall project objectives. Inother words, the emphasis was changed from one ofmaximising e�ciency to one of reporting when an op-eration would be completed and if it was on schedule.With this change to a cellular structure, the line

manager acted as a coach in a supporting role, ratherthan giver of explicit instructions, dealing only withthe overall organisation of the operation into teamsand delegating the rest to subordinates and only help-ing them when requested. There was also more empha-sis on cost reduction and value management. In thisnew environment, the workforce was encouraged tomake their own decisions, seek improved value formoney and organise their own day's work.The structure of the teams themselves was also care-

fully considered before they were formed. The teamleader and project manager together estimated thenumber and skills mix of people required for the teambased on the work required to be done by the team.They then interviewed people in order to gain the cor-rect skills mix and personality traits for the team. The

Figure 1 Changing the organisation of construction teams

Cecentralised site managementÐa case study: P. Barber et al.

114

Page 3: Decentralised site management—a case study

team leader then took responsibility for the team bud-get and completion targets, an overall view of oper-ations and his/her main task was to make sure thateveryone is pulling in the right direction and that theyhave the tools necessary to do the job. Team leaderswere also responsible for producing, in liaison with therest of the team, a three monthly program.The foremen were responsible for overseeing the

weekly running of their part of the works. Their maintask was to make sure the men doing the work havethe tools and equipment to do it. The foremen pro-duced the weekly work schedules in consultation withtheir men and did the preparation so that the workcould be done.The gangermen looked after each individual gang

undertaking a particular task on the weekly work sche-dule. They planned and executed the day to day workof each gang. Their main task was to make sure thatthe gang could do that day's work and that everythingwas set up for the next day's work. An example of theganger-men's new responsibilities was that they weregiven the authority to draw small hand tools and localsupplies from the stores.The engineers, quantity surveyors (QS) and sur-

veyors acted as a support function within the team:making sure that the correct materials are ordered,preparing method statements and procedures, layingout the work, etc. The engineer's role as a manager, asit occurred in the old system, was greatly reduced.This method of working represented a radical depar-

ture from the traditional approach where all the powerand responsibility rested with the engineer in charge ofthat piece of work. In the previous system, the engin-eer would prepare the weekly work schedules andsupervise the day to day operations. This meant thatthe engineers often had very little time to do theirintended work and spent most of their time supervisingthe day to day operations. This shift in power awayfrom the engineers to the foremen caused resentmentamong some of the senior engineers at ®rst. It wasnecessary to provide some training for foremen andgangermen to equip them to meet their new roles.At the earlier stages of the project, the new decen-

tralised project structure and team working seemed tobe producing some improvements in productivity, butonly very slowly. It seemed to take time to build up ane�ective team and get used to operating in a new cul-ture and new ways of operating. There were also someproblems with communications between teams whichmeant that although the work was being done in theteams, they were still often pulling in di�erent direc-tions. This became manifest in the project slippingbehind schedule and over budget. More seemed to beneeded than just delegation and team working. Theproject management had not lost faith with theirdecentralised approach, but they did feel that some-thing additional was needed.

Resolving the problem

With the help of Balfour Beatty's own BusinessImprovement Team, site management identi®ed aselection of problems using the Goldratt ``ThinkingProcess Approach'',4, 5 which is a systematic way ofmapping out a problem, understanding the blockageswhich are preventing the development of a satisfactory

solution and de®ning a series of steps to remove them.After conducting this analysis, the problem areasidenti®ed were:

1. Poor communication: through the increased del-egation, top management often did not know whatwas happening on site. The engineers as the teamleaders now felt that they had an inadequate knowl-edge of what their teams where doing. Moreover,there were still in¯uences of the previous culturewhereby everybody wanted their part of the oper-ation to look good so they covered up the problemsand only reported the good aspects of performance.

2. Poor Information: the poor communications meantthat the information passed on was often wrong orinaccurate.

3. Inaccurate planning: planning was often incorrectbecause the information received was wrong, sowrong assumptions were made as to where the pro-ject was in terms of completion.

4. Training/education issues: the majority of personnelon site where only skilled in one very narrow areasuch as sca�olding, steel ®xing and concrete pour-ing. This meant that a succession of trades wouldhave to come and do their piece of work before anypart of the job was completed. Despite the intent,the teams had not become truly multi-functional.

5. Motivation issues: the bonuses paid to the men werestill based on the old e�ciency-based performancemeasures rather than team performance. This oftenmeant that di�erent trades got in each others wayas they were trying to meet their own speci®c tar-gets.

It became clear from this list that the majority ofthe problems were due to the attitudes that membersof the work force were adopting. They had not com-pletely left behind the previous culture. It was at thisstage that the ``no-blame culture'' was introduced andincreased e�orts made by project management tochange the management style to a more management-by-consensus approach. The no-blame culture was tosupplement the new delegated approach by encoura-ging truthful reporting.

Another key requirement was also revealed: even ifthe transmission of truthful data could be ensured,success would not be achieved if there was not animproved mechanism for overall co-ordination of theproject to replace the form of the previous system ofco-ordination through detailed instructions. Moreover,reverting to tight central control over detail was notan option project management wanted to consider.Hence, such a co-ordination mechanism had both tobe consistent with a decentralised form of operation,and to encourage adaptability and ¯exibility to re-allo-cate resources and schedules to respond to unexpectedevents. The means chosen to do this was a risk man-agement and co-ordination approach based uponGoldratt's Theory of Constraints (hereafter calledTOC).6 This was an approach which had recentlycome to the attention of the Balfour Beatty BusinessImprovement Team.

Cecentralised site managementÐa case study: P. Barber et al.

115

Page 4: Decentralised site management—a case study

Using the theory of constraints approach to riskmanagement

The ®rst task was to review durations for the activitiesrequired to complete the programme. This was nowdone by consulting with the teams who would actuallybe doing these tasks. This process was long and ardu-ous as every activity had to be thoroughly analysedand a realistic time given to it. With these durationsdetermined, the next task that had to be tackled wasthe identi®cation of the key activities that form thecritical path. This task was supposedly easily achievedas the critical path had been identi®ed in the originalproject plan, but the project had never run to this planand this had caused the critical path to move andthese changes had not been monitored through revi-sions of the critical path analysis. Once the criticalpath was revised, a new project plan was drawn upalong TOC lines, but there was one essential and vitaldi�erence from a conventional critical path analysis.Critical path analyses are well known and in wide-

spread use. They show the planned time duration andsequence of activities and help to determine the moste�cient scheme for completion on time. The shortesttime within which the project can be completed is thelength of the critical path and activities on this pathhave no spare time (slack or ¯oat). Activities o� thecritical path have ¯oats of time.When traditional centralised management styles are

used, it would be usual to give directions to subordi-nate managers to complete detailed functional tasksand tell them the expected duration of the task andtheir ¯oat times. This has had the tendency to encou-rage subordinate managers to see themselves as havingindividual time bu�ers equal to their ¯oats for non-critical activities such that no harm was seen in takingup these ¯oats. This then allows non-critical activitiesto become critical, and with the uncertainty involvedin construction, can soon lead to the project fallingbehind schedule. To counter this, the main thrust of theTOC approach is to protect the critical path from beingdelayed by events occurring o� the critical path(Goldratt,6 p. 157).

Under the TOC approach, ¯oats are not allocated toindividual activities. Subordinate managers are giventheir tasks (whether functionally organised or not),with an exact speci®cation of when they are to com-mence the task and the duration it is expected totakeÐno uncertainty allowance is given. An uncer-tainty bu�er is, however, estimated for the whole pro-ject and this will be managed by the project manager.This is called the completion bu�er and it is meant tobe su�cient to cover all the uncertain events whichmay be encountered such that the completion time,including this bu�er, should not be exceeded.For pragmatic management reasons, the total com-

pletion bu�er is broken up into three parts. The ®rstof these is a safety zone: the project can be permittedto go into this zone without any problems. This bu�ersection is used in construction to cope with minordelays caused by, for example, the weather. The sec-ond zone in the bu�er is the planning zone. If the pro-ject enters this zone plans are drawn up to pull it backon schedule. These plans are produced, but unless theproject enters the third zone they are not implemented.The third zone therefore is the implementation zone.The aim of this method of controlling the work is notto enter any of the zones, but to have a total contin-gency with warning, get ready and do somethingaction zones.The total completion bu�er is not the only uncer-

tainty bu�er in a TOC approach. Major projectsusually have separable strings of activities that have acritical path of their own before they feed into themain critical path of the whole project. These stringsare organised in exactly the same way with the bu�erbeing at the end of a string of activities. This bu�er isknown as a feeder bu�er and it is split up in just thesame way as the completion bu�er. This arrangementis the key to protecting the critical path from interfer-ence through delays in the strings (sub-projects) of fee-der activities. This multi-string structure of bu�ers in aTOC based project management system is illustrated inFigure 2.

Figure 2 The use of bu�ers in a TOC Program (Source: The Goldratt Institute)

Cecentralised site managementÐa case study: P. Barber et al.

116

Page 5: Decentralised site management—a case study

It can be argued that the TOC approach does noth-ing more than employ a sensible and fairly traditionaluse of contingency allowances. But to focus upon thatwould be to miss the key point which is that thismethod of planning relies on people sticking to theproject plan and organising themselves and their teamsto do whatever is necessary, using their local knowl-edge and discretion to start on the required date andto achieve at least the required output to ®nish on therequired date. To ensure this happens, all the total pro-ject uncertainty bu�ers remain the property of the pro-ject manager and only he/she is allowed to use them.Similarly, all feeder bu�ers are at the disposal of onlythe relevant ``string managers''. The project manager(string) manager must have the ability to step backand look at the project (string) as a whole and then todecide what, if any, action is required to keep the pro-ject (string) on track. He/she, therefore, responds tothe emerging uncertainties and reallocates resources toe�ect the necessary co-ordination of the whole project(string). The project (string) manager delegates discre-tion on method of work as much as is possible, but istough on delivery of completed task within speci®edtimes and costs. While the project (string) managers'subordinates are aware of the total project uncertaintybu�er, they are not given individual ¯oats for theirspeci®c tasks nor given discretion to use the project(string) bu�er. Recently, the presence of a bu�er (¯oator inventory) has tended to be viewed as a sign ofnon-lean operation, but if subordinate managers andthe work force have no knowledge of ¯oats for theirspeci®c tasks and are made aware that they must notexceed allocated times and costs and, moreover, it istheir task to plan to recover time if they appear to befalling behind schedule, any slackness due to the pre-sence of ¯oats/bu�ers should be avoided. It can alsobe seen now how the TOC approach facilitates a dis-aggregation of the project into sub-projects (strings),so that the bene®ts of decentralised modes of oper-ation can also be obtained, while improving risk man-agement and overall project co-ordination.To implement this management procedure, the pro-

ject manager held weekly meetings with each of themain task teams. In these meetings the progress ofthat team over the week would be discussed and anyactions required would also be discussed and planned.These meeting relied on the teams providing accurateinformation, and not covering up problems asoccurred under the old system of working. The `no-blame' culture helped to encourage this. If con¯ictsoccurred between teams over resources, they would besettled by the project manager allowing the team whohad the work which had the most impact on the pro-gram have the resources ®rst.

Evaluation of the approach

An assessment of the changed management processwas carried out by means of structured interviews witha cross section of the workforce ranging from the pro-ject manager to ganger men, plus subsequent extensivediscussions with the project leader and the internalbusiness improvement consultant. The aim of theseinterviews was to determine if they felt the changes onthe site had improved their work in terms of pro-ductivity and quality. The interviews also attempted to

®nd hard evidence to back up the ®ndings of the inter-views. The ®rst person to be interviewed was the pro-ject manager.

The project manager's views

The project manager and the internal consultant haveworked hard over the past two years implementing thechanges to both the organisation and the culture ofthe site. When TOC was introduced, a 12 month planwas formulated (to end on the 20th December 1996)Over this period a 96% adherence rate was achieved.Before TOC was introduced the site was only mana-ging to achieve 52% adherence to its weekly workschedules.To produce this dramatic change, much e�ort was

needed to change the attitudes of the work force fromthat of confrontational (and only doing what you weretold to do) to a more co-operative consensus drivenmanagement style. There was a drive to produce anopen and honest dialogue between all the people onsite. This was vital for the communication of correctand accurate information, but it took time to build thenecessary level of trustÐit did not arise naturally justbecause teams with more discretion were formulated atthe outset of the project.The project manager stated that the TOC approach

had forced him to step back and look at the project inaggregate terms. The increased delegation allowed himtime to make more strategic decisions and reallocateresources in a more dynamic manner. This, though,could only be made if the information available to himwas correct. The higher quality information, due tothe weekly team meetings, allowed him to make moreinformed decisions. He now had a more reliable viewof how the project stood in relation to the overall pro-ject plan and how the various activities that werebeing undertaken impacted on that plan. Before thesechanges were made, the project manager would knowhow much of various materials had been moved, laid,erected, etc., but not if this was done in the right placeat the right time to meet the overall goal of successfulproject completion.This approach also meant that quality improved

because a better planned and executed job produces abetter quality product. He stated that, in his opinion,the quality of the job had improved and the workdone on gaining the trust of the workforce created animproved working culture. To support this claim, theclient, the Highways Agency, has commented on theexcellent quality of the job.

Team leaders' interviews

The team leaders agreed that TOC and the associatedrevision to team operation had brought about bene®tsin the way they operated. First, they con®rmed thatthe teams were reorganised into logical splits alonggeographical lines within the job and were of a multi-disciplinary natureÐeach team was responsible forbuilding one particular structure and worked on itfrom start to ®nish with only the road laying beingseparate. As an activity applicable throughout thewhole project, it was felt that it still made sense to or-ganise that on a functional basis.

Cecentralised site managementÐa case study: P. Barber et al.

117

Page 6: Decentralised site management—a case study

The team leaders were responsible for the overallstrategic planning for their section of the job and itsexecution. These plans were drawn up with the helpand assistance of the team members involved, so creat-ing a sense of ownership of the plan. The team leader'smain job was to look ahead and plan for the nextcouple of months. The team leader took a long termview of what was being done in their section and theynow had to avoid being drawn into the complexities ofthe day-to-day operations.One of the team leaders stated:

``that it was often di�cult to maintain our distancefrom day to day operations, but it did make the work

go much smoother.''

They also said that the TOC approach had helpedin providing a means of analysing problems as theyoccurred and a logical set of steps to follow to adaptactivities to ensure that the goal was still met. Thealterations in management style also meant that theteam leaders were more involved in decision making atthe top level and the team members were moreinvolved at the team level.The better communications between the members of

the team and the multi-disciplinary nature of theteams meant that problems were often spotted earlieras people were more aware of what would follow onfrom their work. These problems were then dealt withbefore they reached the stage where rework had to bedone and this produced a consequent saving in costsand general improvement in quality of work.All the team leaders were in agreement that the

changes that had taken place were for the better andthat they would like to organise their next job in thesame way. They all felt that these working practiceswould bene®t any civil engineering project. They alsofelt that the changes that had taken place were onlythe start and the new set-up would provide a goodbasis for further development on work in future con-tracts.

Team members' views

A selection of team members were interviewed includ-ing a foreman, section engineers and a quantity sur-veyor. In total 10 people were interviewed mainly fromthe production teams. All persons interviewedexpressed their satisfaction in the new ways of workingand con®rmed that it improved communication withinthe team and led to better working practices, althoughit was stated that communications between teamscould have been yet further improved. It was alsomentioned that progress had been slow and gradualonce the teams were set-up with no achievement ofstep changes in procedures or working practices. Thisslow change meant that many of the respondents haddi�culty actually determining what changes had takenplace and exactly when, but, once they started thinkingabout it, they soon agreed that a large accumulationof changes had occurred over the last three years ofthe project.They stated that the prime example of these changes

is the way in which the team structure was organisedand the work was plannedÐwith the team leader tak-ing an overall view and the foremen planning for thenext week's work and the ganger men organising the

day-to-day operations. They con®rmed that the engin-eers and quantity surveyors had worked as supportfunctions to the actual work, rather than directing it.This meant that the planning of day-to-day activitieswas done much closer to the work face and so it wasbased on improved, up-to-date predictions. The plan-ning was also not done alone by a particular ganger-man or foreman, but with the input from the rest ofthe team. One foreman commented:

``The team does not always make the decisions that Ifeel are right, but I have to go along with them as that

is what the team decided.''

This action alone was said to have lead to theimprovements in program adherence because the pro-grams now represented a realistic picture of what wasactually happening.The respondents were then asked about the quality

issues. They all responded that in their opinions qual-ity on site had improved for a number of reasons:

1. The improved planning meant that work could becompleted in one go without being left for periodsof time because men were pulled o� that particularjob to deal with a piece of work which was behindschedule.

2. The use of the team structure meant that peoplehad a sense of ownership about the work they weredoing and that if their work was poor it woulda�ect another person or gang in their team.

3. The improved communications between team mem-bers meant that problems were often sorted outbefore they became real quality issues and potentialdisasters could be averted.

General observations from visits to sites

The ®rst observation made is that the site appeared tobe in good order and it has a feeling of control with``nobody running around, shouting and screaming''.This impression is further enhanced by the fact thatthe personnel knew what was going on at the site andcould describe how their work ®tted into the overallplan.The sta� all seemed highly motivated and all

focused on getting the job completed on time. This isbest illustrated in the fact that they all provided verysimilar answers to the structured interview questionsand, when asked if they would choose to work thisway again, they all answered that they would do so.Even so, there was still room for further improve-

ment. There were still two main problems thatrequired attention on this site when this study wasundertaken:7

1. Although the previous inherent trait of covering upproblems and incidents had been reduced consider-ably due to the changes implemented on this site,there were still traces of this practice which neededto be eradicated.

2. Even towards the end of the project, the narrowskills base of the workforce still prevented a full re-alisation of the bene®ts of multi-functional oper-ation. This problem occurred throughout theworkforce with site managers and engineers lacking

Cecentralised site managementÐa case study: P. Barber et al.

118

Page 7: Decentralised site management—a case study

management and planning skills, foremen lackingteam building and organisational skills and the menhaving only a very narrow focused skill base suchthat the tasks that they can deal with are limited inrange. Engineers are not exempt from this problem;often they specialise in only one particular ®eld ofoperation.

Conclusions

To conclude a list of the key ®ndings of this researchis o�ered:

. At the end of the project, the site was seen to havebeen a successful one; it was now seen as havingbeen well organised and e�ective.

. The 96% programme and budget adherenceachieved was in sharp contrast to other projects runby the company which often only managed signi®-cantly lower programme adherence with cost overruns on various activities.

. There was also a noticeable improvement in qualityobserved.

. This site organisation also produced a well organ-ised and motivated work force who said that theyhad the tools available to them to do their jobse�ectively.

. These conditions, it was stated, had been so muchbetter than on other sites and the workforce wouldall like to continue working in this fashion.

. The study also showed that the ideas contained inThe Last Planner are worthy of recognition as asimilar system, namely this TOC based system, hasbeen developed and used in practice in BalfourBeatty.8

. From a wider perspective, the case also suggestsvery important hypotheses about the nature of pro-ject planning. Jolivet and Navarre3 emphasise that amove to more decentralised team work and greaterautonomy within the project organisation can onlybe achieved by a thorough change in attitude andculture. This case strongly supported that prop-osition.

. Moreover, the culture had still not fully adapted bythe end of the project and was certainly not univer-sally present throughout the company studied. Thissuggests that one should not expect to be able tochange to a widely held set of meta-values through-out the company on the basis of application, in onlyone project, even a successful one.

. Even within a project, it takes time to train anddevelop sta� in new ways of working. Meanwhile,one must be pragmaticÐpro®ts have to be earnedand projects completed. The case study illustratedhow the process of changing styles of operation hadto be managed within the projectÐone cannot relyon it just being ordained from above.

The current view in the company is that it is prema-ture to decide that this must be the only way to organ-ise all its large projects. Thought is needed as towhether there are not contingent factors which a�ectthe style with which projects are best managedÐper-haps along the lines suggested by Jolivet and Navarre.3

While, in the company, the issue is not yet fullyresolved. It seems quite clear that any claims that ben-e®ts will be automatically derived from delegationshould be viewed with care. There will always be theneed for co-ordination and risk management for theproject as a whole. Advocates of delegated forms ofoperating are not completing their task if they do notalso show how this task of co-ordination is to be per-formed in a way that is consistent with decentralisedproject management. The TOC was the way the man-agers of the A13 project decided to ®ll this gap. It isnot the only approach, but it did prove e�ective in thisinstance.On the other hand, it has been suggested by some in

the company that, if the decentralised teamwork basedapproach had not been attempted initially, TOC andcultural change would not have been required. Ourresearch was not able completely to resolve this issue,but it was able to demonstrate how a TOC and decen-tralised approach can complement each other. Furtherempirical research is needed to identify conditionsunder which construction projects are best run underdecentralised or centralised modes of operation and torelate that to the Jolivet and Navarre3 contingencyframework.

Acknowledgements

The authors gratefully acknowledge the ®nancial sup-port of the EPSRC and Balfour Beatty for The AgileConstruction Programme of which the researchreported here is only a very small part. Thanks arealso due to Balfour Beatty for permitting us to publishthis material in the interests of stimulating a widerdebate on methods of improving construction manage-ment.

References1. Anthony, R., Dearden, J. and Govindarajan, V., Management

Control Systems, (Second edition), Irwin, Homewood, Illinois,1992.

2. Peters, T. and Waterman, R., In Search of Excellence: lessonsfrom America's best-run companies, Harper & Row, New York,1982.

3. Jolivet, F. and Navarre, C., Large-scale projects, self-organisingand meta-rules: towards new forms of management. TheInternational Journal of Project Management, 1996, 14(5), 265±271.

4. Goldratt, E., Theory of Constraints, North River Press, NewYork, 1990.

5. Noreen, E., Smith, D. and Mackay, J., The theory of constraintsand its implications for management accounting, North RiverPress, New York, 1995.

6. Goldratt, E. op.cit. Some we have met in the civil engineeringindustry claim that this approach is little di�erent fromapproaches used a number of years ago. Moreover, it can beargued that it is simply a particular style of using logic trees.Our point is simply that this Balfour Beatty project team cameacross this approach through contact with consultants from theGoldratt Institute at the time that the project leaders werebecoming aware that improved means of co-ordination wasneeded to operate the new management style with success. Theysaw the Goldratt approach as serving these needs. Goldratt hassince published these ideas, see E. Goldratt, Critical Chain,North River press, New York, 1997.

7. This study was undertaken towards the end of the constructionproject.

8. Ballard, G., The Last Planner, North California ConstructionInstitute, Monterey, California, 1994.

Cecentralised site managementÐa case study: P. Barber et al.

119

Page 8: Decentralised site management—a case study

Patrick Barber is a graduate ofCoventry University in Manu-facturing Engineering and heobtained his doctorate at theUniversity of Sunderland forresearch into design management inthe consumer durables industry.Patrick joined the AgileConstruction Initiative in the Schoolof Management of the University ofBath in 1996, where his work isfocusing upon quality issues andimproved management processes formajor construction projects.

Cyril Tomkins is Professor ofBusiness Finance in the School ofManagement, University of Bath.He has published extensively (ap-proximately 100 journal articles,books, Government reports, etc.) ona range of issues related to ®nancialcontrol in large organisations, bothin the private and public sectors.His main interests concern the wayin which management accountingand ®nancial management interfacewith strategic and behaviouralissues. He is one of the three princi-pal holders of the EPSRC/BalfourBeatty grant supporting the Agle Construction Initiative.

Andrew Graves is a Professor inThe School of Management,University of Bath and Director ofboth the School's InternationalAerospace Research Programmeand Agile Construction Initiative.He is also European Director of theMassachusetts Institute ofTechnology International MotorVehicle Programme. After spendingtime in both the Army and Formula1 racing. Andrew has specialised inthe study and implementation oflean production processes andidenti®cation of ``best practice'' andthis paper represents the ®rst published journal article from theAgile Construction Initiative which is attempting to apply such con-cepts across both design and operations within this industry.

Cecentralised site managementÐa case study: P. Barber et al.

120