Kamu sektörü konut projelerinde ihale aşaması planlama sürecine yönelik bir model önerisi
A model proposal for the tender stage planning process in public sector housing projects
- Tez No: 840972
- Danışmanlar: PROF. DR. HAKAN YAMAN
- Tez Türü: Doktora
- Konular: Mimarlık, Architecture
- Anahtar Kelimeler: Kamu inşaatları, Konut yapımı, Proje planlama, Çoklu doğrusal regresyon, Public constructions, Housing contruction, Project planning, Multiple linear regression
- Yıl: 2023
- Dil: Türkçe
- Üniversite: İstanbul Teknik Üniversitesi
- Enstitü: Lisansüstü Eğitim Enstitüsü
- Ana Bilim Dalı: Mimarlık Ana Bilim Dalı
- Bilim Dalı: Proje ve Yapım Yönetimi Bilim Dalı
- Sayfa Sayısı: Belirtilmemiş.
Özet
İnşaat projelerinin ihale evresinde planlama ve programlama faaliyetlerinin etkinliği, projelerin bir sonraki yapım aşamasında meydana gelebilecek sorunların ve risklerin asgari düzeye indirilmesini sağlamaktadır. Türkiye'de özellikle kamu inşaat sektöründe gerçekleştirilen ihale süreçlerinde planlama ve programlama çalışmalarına gereken önemin verilmediği görülmektedir. İhale evresindeki planlama ve programlamadaki bu yetersizlik nedeniyle yürütülen projeler ile teslim süreleri ve yatırım kararları arasında uyumlu olmayan nitelikler görüldüğü açıktır. Bu sorunun çözülmesi için öncelikle kamu sektöründeki inşaat projelerinin taahhüt edilen zamanda teslim edilmesini sağlayarak işveren ve hak sahipleri arasındaki anlaşmazlıkları önlemek ve projelerin Tahmini İş Süresini belirleyip işin zamanında bitirilmesini sağlamak önem arz etmektedir. Böylelikle projelerde hem kamunun hem de hak sahiplerinin parasal ve zamansal kayıplarına engel olunması mümkün olacaktır. Türkiye'deki kamu inşaat sektöründe en aktif ve büyük kurumun TOKİ olması nedeniyle bu doktora tezinde TOKİ tarafından yürütülen konut projelerinde gecikme sürelerinin belirlenmesi ve belirlenen bu gecikme sürelerini azaltacak Tahmini İş Süresini hesaplayacak bir modelin önerilmesi amaçlanmıştır. Doktora tezinde karma araştırma yöntemi kullanılmıştır. Öncelikle TOKİ yetkilileri ile yapılan görüşmeleri içeren bir vaka çalışması metodolojisi benimsemiştir. Görüşmeler sırasında mevcut 3500 TOKİ projesine ilişkin ayrıntılı bilgileri gösteren bir rapor elde edilmiştir. Görüşmeler ayrıca TOKİ'nin mevcut inşaat süresini hesaplamak için konut sayısı, çizim türü ve çalışılmayan günler olmak üzere üç faktörü kullandığını göstermiştir. Çalışmanın bulguları, TOKİ tarafından hesaplanan toplam inşaat sürelerinin, resmi yüklenici sözleşmelerine keşif süresi olarak aktarılırken yüksek düzeyde kısaltıldığı görülmüştür. Bunun sonucunda TOKİ konut projelerinde birçok gecikmenin ortaya çıktığı tespit edilmiştir. Sonuç olarak toplam inşaat süresinin daha hassas ve pratik bir şekilde tahmin edilmesi için süreyi etkileyen ek faktörleri kullanarak Tahmini İş Süresine ulaşan yeni bir modele gereksinim olduğu saptanmıştır. Bu sebeple doktora tezinin ikinci kısmında bir model önerine yer verilmiştir. Mevcut 2800 adet TOKİ inşaat projeleri içinde %25,71 (1530) oranla en fazla konut projelerinin geciktiği ve konut projelerinin de kendi içerisinde %47,06 (720) oranında gecikmeye sahip olduğu belirlenmiştir. Bu nedenle, modelin yalnızca konut projelerine uygulanmasına karar verilmiştir. Doktora tezinin bu bölümünün temel amacı,“Tahmini İş Süresi”olarak adlandırılan toplam inşaat süresini daha doğru tahmin etmek için literatür taraması, TOKİ yetkilileri ile görüşmeler ve TOKİ'den elde edilen belgelerden elde edilen toplam 11 adet faktörü içeren yeni bir model önermektir. Önerilen model, SPSS 26.0 yazılımı ve istatistiksel yöntemler kullanılarak Temel İş Süresini oluşturan ilk üç faktöre diğer sekiz faktörün etkisiyle belirli değerlendirme ölçütleri doğrultusunda Tahmini İş Süresini hesaplamaktadır. Veri analizi sırasıyla çoklu regresyon analizi, CHAID ve CART ile yapılmıştır. Bulgular %95 güven aralığına göre yorumlanmıştır (p<0,05). Çalışmanın bulguları regresyon, CHAID ve CART analizi için sırasıyla on bir faktörden sekizinin, yedisinin ve beşinin anlamlı olduğu ve Temel İş Süresini etkilediklerini göstermiştir. Kestirimler ve standart hatalar, üç istatistiksel yöntemin tümünün geçerliliğini kontrol etmek ve doğrulamak için hesaplanmış ve her üç yöntem için bulunan sonuçlar kendi aralarında karşılaştırılmıştır. Sonrasında model test edilmiş ve regresyon formülünün geçerliliği gösterilmiştir. Bulgulara göre temel iş süresi(F1+F2+F3), proje önceliği (F4), projenin karmaşıklığı (F5), projenin zorluk derecesi (F7), projenin finansal riski (F8) ve proje bölgesinin iklim koşulları (F10) Tahmini İş Süresini önemli ölçüde etkilemektedir. Temel iş süresi, Tahmini İş Süresi üzerinde en yüksek etkiye sahiptir. CHAID ve CART, standart sapmalar açısından regresyona göre daha iyi performans göstererek, daha karmaşık değerlendirme kriterlerine sahip olan F9, F10 ve F11 faktörleri nedeniyle hesaplama yönteminin tahmin yeteneğini artırmıştır. Modelin uygulama safhasında öncelikle TOKİ'nin mevcut Temel İş Süresi hesabına ilişkin değerlendirme ve eleştiriler yapılmıştır. Bu noktada TOKİ'nin sözleşme sürelerini belirlerken hesapladığı iş sürelerini dikkate değer miktarda kısaltarak sözleşme sürelerini oluşturduğu gözlemlenmiştir. Sonrasında her bir yöntemde bulunan anlamlı faktörler kullanılarak 1530 konut projesi için Tahmini İş Süreleri ve geciken konut proje sayıları yeniden hesaplanmıştır. Modelin uygulanması öncesi, mevcut duruma göre geciken konut proje sayısında belirgin bir azalma olduğu gözlemlenmiştir. Modelin uygulanması öncesinde geciken proje sayısı 720 olup, 1530 proje içinde %47,06 orana karşılık gelmektedir. Model uygulandıktan sonra geciken proje sayısının regresyon yönteminde %22,88'e (350 proje), CHAID yönteminde %18,63'e (285 proje) ve CART yönteminde ise %19,54'e (299 proje) düştüğü görülmüştür. Sonuçlar önerilen modelin geciken konut proje sayısını azaltmayı başardığını göstermektedir. Sonuç olarak bulgular, önerilen modelin konut projeleri için ideal süreyi tahmin etmek için geçerli ve güvenilir bir araç olduğunu desteklemiştir.
Özet (Çeviri)
The effectiveness of planning and programming activities in the tender phase of construction projects ensures that the problems and risks that may occur in the next construction phase of the projects are minimized. Although scheduling at the tender stage is regulated by relevant laws and authorized institutions, there is no such regulation for the public construction sector in Turkey. This is because only the lowest bid price is given priority without planning during the tender stage. When choosing a contractor, not only low costs but also construction duration should be taken into consideration. Moreover, there is a lack of literature on the calculation of construction duration at the tender stage for housing projects. In order to solve this problem, it is important to prevent disputes between employers and beneficiaries by ensuring that the construction projects in public sector are delivered on time and by determining the estimated construction duration of the projects and to ensure it is completed on time. Thus, it will be possible to prevent time and financial losses of both the public and beneficiaries of the projects. Since TOKI has been the most active and largest institution in the public construction sector in Turkey, the purpose of this doctoral thesis is to determine the delays in the housing projects carried out by TOKI and to propose a model that calculates the Estimated Construction Duration by reducing these delays. Two main problems for this research are: (1)“What are the significant factors affecting the total construction time in the proposed model for the tender stage planning process?”and (2)“What is the effect of this model on the delays in existing mass housing projects?”This doctoral thesis used mixed methods design. The research adopts a case study methodology involving interviews with TOKI officials. A report showing detailed and updated project information about 3500 TOKI projects was obtained during the interviews. Interviews also showed that TOKI has been using three factors which are number of houses, drawing type, and non-working days to calculate baseline construction duration. Findings of the study have shown that calculated construction durations by TOKI were way higher than of those indicated in official construction contracts. Comparison between construction durations by using TOKI's own calculation method and official construction contracts have revealed that TOKI was significantly shortened the durations while deciding official construction durations which ultimately led to many delays in the housing projects. In conclusion, it was suggested that a new model reaching estimated construction duration by using additional factors affecting housing projects' duration was necessary to predict total construction duration more reliably and practically. Therefore, the second part of the study involved a model proposal. Descriptive statistics showed that number of projects with delays among completed TOKI projects (2800) is the highest for housing projects (1530) by 25.71% while the number of housing projects with delays among all housing projects was 720 (47.06%). Therefore, the model is applied only for housing projects. The main objective of this part of doctoral thesis is to propose a novel model composed of eleven factors chosen from the literature review, TOKI interviews, and TOKI documents to estimate more accurate total construction duration which was named hereby“Estimated Construction Duration”. Literature has shown that various factors affect construction duration of housing projects such as the type of project and tender, the complexity of the project design, construction volume and height (number of floors), site characteristics, weather conditions and the geographical location of the project, project characteristics and pre-construction scheduling, financial conditions, seismicity, and supply conditions. 56 factors addressed in the literature regarding construction duration and they are reduced to three factors for proposed model by considering force majeure, tender stage, attribution to the employer, eligibility and applicability to statistical methods. The proposed model considers baseline duration which utilizes three factors and calculates estimated duration by using certain assessment criterion for each additional eight from the literature (in total eleven) factors by using SPSS 26.0 software and statistical methods. Quantitative methodology involves development, validation, testing, and implementation of proposed model. Statistical data analysis was performed using multiple linear regression analysis, Chi-squared Automatic Interaction Detection (CHAID), and Classification and Regression Tree (CART) methods. Multiple Linear Regression Analysis with the backward elimination method and non-parametric CHAID and CART analyses were used to determine the factors affecting the estimated construction duration. Multiple Linear Regression Analysis is a statistical method that simulates the causality relationship between more than one independent variable and illustrates the extent to which the dependent variable is explained by the independent variables. CHAID is used to identify and analyze the classified dependent variables. The purpose of this analysis method, which is frequently used in data mining, is to divide the dataset, dependent variables and independent variables used in the analysis into subcategories that are more homogeneous. It was observed that the reliability and accuracy of the analysis results depend on the division of the dataset into homogeneous subcategories. CART is a non-parametric statistical method used to estimate categorical and continuous dependent variables. Depending on whether the dependent variable is continuous or discrete, CART provides regression or classification trees. The decision tree is obtained by categorizing the independent variables that affect the dependent variable into binary subgroups according to the interactions between the variables. Repetitive binary subgrouping continues until decision points are reached. In regression analysis, the significance of the independent variables is evaluated using numerical rather than categorical variables. In contrast, the CHAID and CART methods introduce variables as decision trees instead of equations. Therefore, in this study, three different statistical analysis methods were used (one equation and two decision trees). These three methods were used to determine estimated construction duration to reach an optimal solution. Validation for regression, CHAID, and CART methods was done by enter and stepwise methods, and 10-fold cross-validation, respectively. Findings were interpreted according to confidence interval of 95% (p<0,05). Analysis have shown that eight, seven, and five factors out of eleven were significant and affected the Baseline Construction Duration for regression, CHAID, and CART analysis, respectively. The findings showed that baseline construction duration (F1+F2+F3), priority of the project (F4), complexity of the project (F5), difficulty of the project (F7), financial risk of the project (F8), and climatic conditions of the project region (F10) have significantly affected the construction duration. The baseline construction duration had the highest impact on the estimated construction duration. CHAID and CART performed better than regression in terms of standard deviations and therefore increasing the predictive ability of the calculation method due to F9, F10, and F11 which have more complex evaluation criteria. The cutoffs and standard errors were calculated to test the validity of all three statistical methods. Enter and Stepwise methods were used for the validity of regression method while 10-fold cross validation, training dataset (70%) and test dataset (30%) were used to find the validity of CHAID and CART methods. The classification accuracy was 62.2% compared to normal construction work end periods. This ratio was calculated as 91.8% for regression analysis, 93.7% for CHAID, and 93.9% for CART. The regression formula indicated statistical significance when the calculation method was tested. The implementation of the regression method for test data (40 housing projects with delays) significantly reduced the number of delayed projects. Testing of the regression method reduced the number of delayed housing projects by 42.50% as the number of delayed housing projects decreased from 40 to 23. The estimated construction duration after testing model was significantly above the contract periods specified by TOKI in the construction contracts. In addition, it was determined that the estimated construction duration obtained by running the model was generally slightly higher than baseline construction duration. This is supported by graphical analysis in which the trend lines of estimated construction duration and baseline construction duration were parallel to each other. In the implementation phase of the model firstly evaluation and criticism were done for TOKI's calculation method for Baseline Construction Duration. It was determined that TOKI was significantly shortening the durations in official contracts. By using significant factors for each method, estimated construction duration for housing projects has been calculated for 1530 projects again and number of delayed housing project were recalculated. It was observed that the number of delayed housing projects were decreased prominently compared to the existing case before the model implementation. The number of projects delayed before the implementation of model was 720 which corresponds to 47,06% among 1530 projects. After the model was applied, it was observed that the number of delayed projects decreased to 22.88% (350 projects) for the regression model while it was decreased to 18,63% (285 projects) for CHAID and 19,54% (299 projects) for CART models. These results show that the proposed model was achieved to reduce delays in housing projects. In conclusion, the findings supported that proposed model is a valid and reliable tool to determine estimated construction duration for housing projects. Since there was no research on the use of CHAID and CART in the estimation of the construction duration of the housing projects, both literature on other topics with similar methods and different methods on the similar construction research were discussed. In similar studies on construction projects, it was found that the estimation was generally made with regression and gave good results. In addition, CHAID and CART methods have shown that they have a good accuracy in estimating the defects that have indirect impact on the completion time of public construction projects and exceeding the desired completion time. In other construction researches, alternative methods are used to predict duration such as Delphi process, mathematical models of integer linear programming, artificial neural networks, sensitivity analysis, and Markov chain models. There are some limitations in this research. First, the estimation of the construction duration includes only mass housing projects, as the majority of public construction projects in Turkey are housing projects that lead to major delays. Although the construction oriented economy in Turkey has been adopted and the construction boom is mainly due to the housing projects of TOKI, a large number of lawsuits filed for housing projects with delays emphasize the ongoing lack of time planning. In addition, other projects such as highways, bridges, tunnels and airports are given relatively more importance for political reasons. The second limitation is that the proposed method is limited to the accuracy of the public construction project data and that the model can affect the estimation accuracy. Third, the main factors found to have a significant impact on the construction duration in the previous studies were partially included in the pre-construction (tender) stage of many factors. For this reason, only the factors affecting the construction duration during the tender stage were considered in this doctoral thesis.
Benzer Tezler
- Yap-işlet-devret modeli inşaat projelerinde eş zamanlı gecikmeler: Türkiye'den bir vaka analizi
Concurrent delays in build-operate-transfer model construction projects: A case study from Turkey
ÇAĞDAŞ ÇOLAK
Yüksek Lisans
Türkçe
2024
Mimarlıkİstanbul Teknik ÜniversitesiMimarlık Ana Bilim Dalı
DOÇ. DR. FATMA PINAR ÇAKMAK
- The bid/no-bid framework utilizing Node2Vec for international contractors
Uluslararası yükleniciler için Node2Vec ile teklif/red çerçevesi oluşturulması
HALİT FERDİ AVCI
Yüksek Lisans
İngilizce
2024
İnşaat Mühendisliğiİstanbul Teknik Üniversitesiİnşaat Mühendisliği Ana Bilim Dalı
DOÇ. DR. ONUR BEHZAT TOKDEMİR
- Burdur ili mermer sektörünün kurumsal ve ekonomik yapısı
İnstitutional and economic structure of marble sector in burdur
AHMET SARITAŞ