Επίδραση της παραγωγικότητας των πόρων στον προγραμματισμό έργων πολιτικού μηχανικού

Abstract

The scheduling and cost management of civil engineering projects requires the realistic determination of theexpected productivity for the deployed resources. The productivity values can be derived from the estimator’spersonal experience, published estimation methods and handbooks or a company’s historical record. Thetransformation of theoretical or expected productivity is accomplished through the use of specific adjustmentcoefficients or “efficiency multipliers”. These adjustment coefficients represent the direct effect of operationalconditions (e.g. working method) or specific productivity factors (e.g. learning rate) on the actual or effectiveproductivity achieved on-site and have been the focus of many research efforts in the past. However, theestimation process becomes even more complicated when the activities under study are not associated with anyknown estimation method. The latter may be the case when activities are executed for the first time or knownactivities are executed under a new operational framework which diverges from the standard method statement.In these situations the major challenge is for the estimation process to be based on a valid procedural frameworkthat will enable the incorporation of new operational factors which are not known a priori, but rather determinedby the estimator.In view of the aforementioned findings, the reliable estimation of productivity and its respective efficiencymultipliers is still an open research area, due to four main reasons: First, there is no structured approach for theirestimation. Productivity is often estimated on an ad-hoc basis, depending on the project characteristics and theunique estimation objectives. As a result, the quantitative relationship between the affecting factors andproductivity is not well understood. Furthermore, the type and values’ range of the efficiency multipliers isassociated with the adopted estimation method. However, it has been concluded that many known estimationmethods, such as those prescribed by equipment manufacturers (e.g. Caterpillar, Komatsu, Liebherr) are noteasily modifiable, since they comprise computational difficulties, while, quite often, the theoretical assumptionsof one method contradict the theoretical assumptions of another similar method for a given activity type andresources. In addition, it should be noted that the efficiency multipliers of these productivity estimation methodshave been openly criticized for yielding unrealistically optimistic results.The second challenge relates to the lack of a contextual framework for the comparative evaluation of the derivedefficiency multipliers. This has led to the provision of fragmented results and the establishment of productionrates which are based on an unknown scope and, hence, are of questionable practical value. Due to their limitedreliability, the yielded productivity values are hardly applicable for accurate construction time estimation. Inaddition, the comparative evaluation of the efficiency multipliers for a specific activity and also across differentactivities or even projects is hindered, since the estimator is not able to evaluate whether the contextualframework, which the multipliers stem from, is the same. Moreover, exact contextual information is needed by the estimators, so as to comprehend the conditions under which specific production rates were achieved, andconsequently, decide which production data would represent the estimated activities in the most realistic way.The third challenge reflects the reluctance of the construction industry to adopt more advanced analyticalmethods for productivity estimation, such as the stochastic analysis with the use of simulation techniques,despite the general consensus indicating that the deterministic analysis fails to grasp the dynamically changingproduction rate of the contemporary, complex civil engineering projects. This is to be attributed to the fact that,despite its obvious advantages, simulation has been often criticized for being more of a “black art”, which yieldsmisleading or non-applicable results. As such, construction practitioners and the industry at large seem to remainunconvinced of the merits stemming from the application of simulation for modeling and analyzing constructionoperations. It has been argued that some of the key reasons for doubting the effectiveness of simulation as adecision tool are the failure to adequately demonstrate the effect of critical physical factors that directly affectthe system’s performance in a clear and understandable manner, the reporting of simulation models that lackbasic information which would enable the repeatability of the experiments for verification purposes and theinterpretation of simulation results without delineating the contextual framework within which the study wasconducted. The latter is especially important in the project planning phase, since the contextual background ofkey productivity or cost data determines the scope of their applicability in the estimating process. Understandingand interpreting statistical simulation results which are not explicitly associated with a specific operationalsetting, especially when the system under study is a complex one, can be a very difficult task.The fourth reason regards the inability of integrating complex productivity factors in the scheduling and costmanagement of projects, based on their dynamically changing values range. A characteristic example is thehuman factor, which expresses the ability and competence of the crews that are deployed for a specific activity.It has been proven that, in the case of complex activities, the human factor reflects the learning rate. However,the learning curves theory is limited to the retrospective evaluation of productivity through the study of historicaldata for completed activities. Therefore, there is a need to implement stochastic methods, such as simulation, inorder to both evaluate historical data, as well as predict the expected on-site productivity.This thesis’ response to the aforementioned research problems is threefold: First, a methodological framework isintroduced that enables the comparative evaluation of fourteen (14) known productivity estimation methods andtheir respective adjustment coefficients. The suggested framework builds upon the classic operations analysistheory and extends its theoretical fundaments. It introduces the concept of “baseline reference conditions” for thedetermination of the operational conditions under which the effect of a specific productivity factor can beneglected. Thus, it allows the transformation of the computational process of a specific productivity estimationmethod into mathematically equivalent operational scenarios irrespective of the estimation method type. Themethodological framework is implemented for the comparative evaluation of earthworks productivity, with aparticular focus on the hydraulic excavator. Secondly, an additional methodological framework is formulated for the comparative evaluation of unknown ornot explicitly defined productivity estimation methods, as well as its respective productivity factors, so as todetermine the adjustment coefficients and define the baseline reference conditions for each parameter. Thesuggested framework is implemented for the creation of statistical models and artificial neural network models,whereas its applicability has been evaluated through a case study for the construction of heavy-type, reinforcedconcrete pavements in a major infrastructure project.Thirdly, this thesis describes the design, implementation and evaluation of a stochastic productivity model and arespective prototype simulation system named CaissonSim, which allows the stochastic analysis of constructionproductivity and cost for the production of floating caissons through the use of discrete-event simulationtechniques. The stochastic model’s structure, its main functional characteristics, as well as the results of its casestudyimplementation in a large-scale marine project are analytically described. The main innovative elements ofthe stochastic model and the simulation platform is the integrated library of simulation model templates that havenot been published before, the ability of simultaneous statistical processing for every possible combination ofproductivity factors for the operational scenarios under study, as well as the creation of dynamic reports for thesimulation experiments providing access to the statistical data of every iteration and system parameter. Finally,CaissonSim has been extended, so as to include the integration of the learning rate of every activity contained inthe simulation model. CaissonSim enables not only the simulation-based assessment but also the prediction ofconstruction productivity as well, which has not been attempted in published literature before.The research findings are categorized in two main research areas:(a) The formulation of a methodological framework which associates, in a statistically valid manner, differentproductivity estimation methods based on known productivity factors and contributes in the measurement andcomparative evaluation of unknown productivity factors with the use of deterministic methods and(b) The implementation of discrete-event simulation, as a stochastic method and a prototype simulation platformfor the analysis of construction activities which have not been scrutinized before for the assessment andprediction of construction productivity.The contribution of the thesis in the first research area is summarized below:1. Contribution to the theory of Operations Analysis through the formulation of a methodology for theanalysis and comparative evaluation of known productivity estimation methods, which enables theevaluation of the reliability and applicability scope for the yielded results.2. The concept of the “baseline reference conditions” has been introduced, which enables the conceptualtransformation of any construction scenario to equivalent mathematical models, so as to enable thecomparative evaluation of different productivity estimation methods in a statistically valid way.3. Formulation of a methodology for the on-site measurement of productivity adjustment coefficientsirrespective from the activity type, the number of productivity factors and the nature of the works. Theonly prerequisite is the mathematical formulation of the modeled construction method through theadoption of the factor model.The contribution of the thesis in the second research area is summarized below:1. CaissonSim simulation system aims at studying and simulating key cost and time parameters for thecaissons production process. There has not been any similar simulation framework published inliterature.2. The system’s functional modules allow for the concurrent elaboration of alternative constructionoperations of a given activity by the use of the scenario implementation generator.3. CaissonSim’s reporting modules allow dynamic access to all simulation iterations at the lowestabstraction level. Thus, the user gains full control of the modeling process and may evaluate all basicparameters contained in the simulation models library.4. The thesis has examined the development of the learning phenomenon for the construction of caissonsby assessing and predicting the expected productivity through the implementation of simulation-basedanalysis. There has not been any similar approach in the past. The comparative evaluation of theresearch is based both on the statistical and the simulation-based approach.Conclusively, the overall research purpose is the creation of a unified system for the estimation, monitoring andassessment of construction productivity by the use of all available analysis techniques (deterministic, stochastic)in order to integrate the productivity estimation algorithms, as they were formulated within the framework of thethesis. Such an approach would result in the creation of an information systems infrastructure which will providereal time information and aid construction operatives in the establishment of a reliable and valid decision makingenvironment.

Το αντικείµενο της διατριβής αφορά στη διερεύνηση της παραγωγικότητας για τον προγραµµατισµό των έργωνπολιτικού µηχανικού µε προσδιοριστικές και στοχαστικές µεθόδους. Οι τιµές της παραγωγικότητας προκύπτουνείτε από την προσωπική εµπειρία του κοστολόγου µηχανικού, είτε από εκτιµητικές µεθόδους της σχετικήςβιβλιογραφίας, είτε από το ιστορικό αρχείο µίας εταιρείας. Η προσαρµογή της θεωρητικής παραγωγικότητας(Qth) στις συνθήκες του έργου γίνεται µέσα από τη χρήση συντελεστών παραγωγικότητας και των αντίστοιχωνδιορθωτικών πολλαπλασιαστών τους µε σκοπό τον υπολογισµό της αναµενόµενης παραγωγικότητας (Qeff). Οιδιορθωτικοί πολλαπλασιαστές εκφράζουν την επιρροή των λειτουργικών συνθηκών (π.χ. µέθοδος εργασίας) ήάλλων µεταβλητών (π.χ. ρυθµός µάθησης εργασίας) στην παραγωγικότητα που επιτυγχάνεται επί τόπου τουέργου και ο ακριβής προσδιορισµός τους έχει απασχολήσει πολλές ερευνητικές προσπάθειες στο παρελθόν.Η διατριβή συµβάλλει στη θεωρία της Λειτουργικής Ανάλυσης µε προσδιοριστικές µεθόδους µέσω τηςσυγκριτικής αξιολόγησης δεκατεσσάρων (14) γνωστών µεθόδων εκτίµησης παραγωγικότητας, έτσι ώστε ναεκτιµηθεί η αξιοπιστία τους και το πεδίο εφαρµογής των αποτελεσµάτων τους. Ακόµη, διατυπώνεται η έννοιατων «συνθηκών βάσης αναφοράς» (ΣΒΑ) που επιτρέπει την στατιστικά έγκυρη «αποµόνωση» των συντελεστώνπαραγωγικότητας που δεν επηρεάζουν την ανάλυση, έτσι ώστε να είναι δυνατή η συγκριτική αξιολόγησηδιαφορετικών µεθόδων εκτίµησης της παραγωγικότητας. Επιπρόσθετα διατυπώνεται µεθοδολογία για την µέτρηση επί τόπου του έργου των διορθωτικών πολλαπλασιαστών της παραγωγικότητας ανεξάρτητα από τον τύπο των δραστηριοτήτων, το πλήθος των συντελεστών παραγωγικότητας και τη φύση των εργασιών.Η συµβολή της διατριβής όσον αφορά στη διερεύνηση της παραγωγικότητας µε στοχαστικές µεθόδους αφορά στη µελέτη και στην προσοµοίωση βασικών κοστολογικών και χρονικών παραµέτρων της διαδικασίαςκατασκευής κυψελωτών κιβωτίων από οπλισµένο σκυρόδεµα (caisson). Αναπτύχθηκε πρωτότυπο λογισµικό προσοµοίωσης (CaissonSim), ενώ αντίστοιχο πλαίσιο προσοµοίωσης δεν έχει παρουσιαστεί στην βιβλιογραφία µέχρι σήµερα. Ακόµη, διερευνήθηκε ο συντελεστής ανθρώπινου παράγοντα µέσω της εξέλιξης του φαινοµένουτης µάθησης για την κατασκευή κυψελωτών κιβωτίων από οπλισµένο σκυρόδεµα και εφαρµογή τηςπροσοµοίωσης για την εκτίµηση της αναµενόµενης παραγωγικότητας. Η συγκριτική αξιολόγηση τωναποτελεσµάτων έγινε µεταξύ της στατιστικής µεθόδου και της προσοµοίωσης. Είναι η πρώτη φορά που πραγµατοποιείται µία τέτοια ερευνητική προσπάθεια.Συµπερασµατικά, απώτερος σκοπός είναι η δηµιουργία ενός ενοποιηµένου συστήµατος εκτίµησης,παρακολούθησης και αποτίµησης της παραγωγικότητας µε τη χρήση όλων των διαθέσιµων τεχνικών ανάλυσης(προσδιοριστικές, στοχαστικές) και µε ενσωµάτωση των αλγορίθµων υπολογισµού των µεθόδωνπαραγωγικότητας, όπως αυτές διατυπώθηκαν στο πλαίσιο της διατριβής.