New model for construction procurement automation

Creative Construction Conference 2014 New model for construction procurement automation David Dudáš a*, Petr Dlask a a Department of Construction Management and Economics, Faculty of civil engineering, Czech technical university in Prague, Thákurova 7, Prague 166 29, Czech Republic Abstract Many of construction engineers realize an urgent need for bringing construction processes into the information age of the twenty-first century. Most of the innovators in this area feel that the platform of Building Information Modeling is the future of construction. However, BIM approach places great demands on technical skills of all construction project participants. This is the reason why BIM is applied at first on so-called high-end projects (expensive and technologically demanding).however, all other construction projects with all participants are also a part of the 21st century, the age of information technology and they should benefit from it. This paper presents principles and methodology of more effective work in a relationship project owner construction economist contractor, with a focus on low-end projects (family houses, etc.). This sector of the construction industry is neglected and not considered as the leader of technological progress. And yet, this work shows, with an example of automation in tender estimates comparison, that the integration of new technologies can start right in this sector. The paper deals with the principles that are necessary for the work of construction economists in their relation to contractors and project owners, with a goal to create an equivalent electronic system for automated use. This is accompanied by a comparison of the costs needed for education and qualification of construction economists for the purpose of the construction cost estimates, with costs needed for the development of new automated system for the same purpose. To confirm correctness of presented principles, a case study is included. Keywords: Automation, construction economics, tender estimates. 1. Introduction Paper is focused on using automation in processes within construction industry. A lot of routine especially administrative work is still done manually by human workers. This creates high number of necessary costs and decreases job satisfaction. Automation alternatives are created only by rich companies for complex projects. The underlying idea is that there is a place for automation carried out by information technologies in routine engineering work, and that such automation could be applied in all project sizes. In the beginning, we will analyze the work of engineers and break it down to pieces which could be alternatively done with computing power using information technologies. Then diverse computer replacements will also be presented for workers activities. Then we will put together small pieces of alternative IT approach and create a tool an application which could be used for automation during the whole process. A comparison is made between cooperation applications (specifically BIM) and applications with the goal of automation. This paper notes the fact that automation can save costs and save time of engineers so that they can focus on more important tasks. Goal of this paper in short term is to point out advantages of full individual process automation, collective knowledge, and analysis of construction engineers work responsibilities and workload. 2. BIM When we talk about using computer science or information technologies in the field of construction industry, we definitely encounter BIM Building Information Modeling (Eastman, 2011). Building Information Modeling system was designed to create a small revolution in the construction project planning, realization and life cycle. The idea of BIM is quite simple. Architect creates a 3D design in BIM compatible software. Design is not a simple drawing. It is created with objects (representing e.g. wall, window etc.) which contain a variety of descriptive information. Objects and software are created expertly to help architects. They do so by preventing obvious errors (e.g. the fact that windows cannot be placed in the middle of a room). BIM system also allows cooperation with other professionals. Construction engineers can do their bit in the design regarding building * Corresponding author. Tel.: +420 603 295 212. E-mail address:david.dudas@fsv.cvut.cz 169

equipment and statics without interfering with the work of the architect. When all this work concerning the creation of a design is finished, the appropriate moment for quantity surveying, time planning and bidding comes. This should be made much easier than with the aid of standard design (drawings). Because design created from intelligent objects already contains information about time, price, quantity, material etc. Professionals agree that BIM is a more efficient way of handling projects than the standard way. However, thanks to its concept BIM has aspects which could be considered as disadvantages. These aspects are noted in this paper, and described to offer comparison and contrast between the traditional method which relies to a high degree on human work, and the proposed method of automation. The most common disadvantages of BIM automation are the following: Project is created by the means of one platform so all participants must be familiar with it. It creates a pressure on special education, training and investments for every company involved. BIM system is merely a tool. It is created for cooperation and more efficient way of projects design and following work. There is still a need for people to work with it. Thanks to complexity of BIM system and to its high cost, only high end projects are being chosen for BIM implementation so far. Thanks to these aspects it is very unlikely that BIM platform will spread among all professional in industry during next decade. And even if it does, it may not solve all problems related to construction documentation. 3. Documentation flow and automation There are at least three main parties in every construction project: architect representing the design, contractor representing a working power to realize the design and the owner representing the capital invested. Construction project itself is a very complex issue which requires cooperation and communication between parties involved. Important communication is realized by documentation flow e.g. blueprints, bills of quantities, bidding documents etc. (Fewings, 2012). Even if the transfer of these documents can be managed by certain cooperation system or their generating can be easier with a sophisticated modelling system (e.g. BIM), there still has to be a human worker to process it. It brings up an opportunity for automation. First of all there is need to analyze engineering work. Such analysis could be automated. It needs to be said that only routine and non-creative work activities are suitable for automation. It follows from current capabilities of computers (Harel & Feldman, 2004). 4. Specifics of engineering work The scope of an engineer s workload is a combination of complex tasks which can be generally divided into two groups: technical engineering work and socially collaborative work (Robinson, 2012). Socially collaborative work includes obvious activities of engineer s workload: communication, collaboration, meeting, collective decision making, collective problem solving etc. These activities have certain common features. They usually involve colleagues and the outputs of collaborative work have a creative component. This is the place for cooperative information systems or BIM systems which are design for socially collaborative type of work. On the other hand there is the part of engineer s work which could be summed up under the heading of technical engineering work activities. This work description includes several general activities: receiving information, offering information, searching for information, locating information source, locating information within source, understanding information, problem solving, decision making, others (Robinson, 2010). According to several sources, we can say that technical engineering activities represent approximately 50% of workers time (Robinson, 2012; McCabe & Wissler, 2010). Activities including human interaction are very difficult to be managed by automated processes. However, technical engineering work activities are quite suitable for automation, they only need to be divided into subgroups regarding the content of creativity and routine. According to the definition of creative activities (Parkhurst, 2011), various activities fall within the structure of 3 base categories: strictly routine activities, strictly creative activities and semi-creative activities (activities with mixed nature. For calculation purposes, we assume they are mixed in 50/50 ratio.). Table 1 presents these tree groups with percentages of average time consumption of engineer s work based on research by Robinson (Robinson, 2010). 170

Table 1. Activities divided into groups according to the creative aspect including percentage of time consumption of technical engineering workload. strictly routine activities semi-creative activities strictly creative activities receiving information 7,22% searching for information 5,85% locating information source 5,75% giving information 9,31% locating information within source 16,67% understanding information 16,10% decision making 9,46% problem solving 14,82% others 14,81% Table 1 offers simple results to our reasoning. Non-creative activities on average represent 20% of total workload and therefore 8 hours of a standard 40-hour work week. These activities are worth an effort for automation. All of the aforementioned general non-creative and semi-creative activities are composed by basic elements according to WBS - work breakdown structure (Haugan, 2001). There are some examples: physical documents searching (electronic and paper), logical decision making, logical analysis, learning, reporting results, making notes, remembering. Detailed level of basic elements depends on situation. There is an example of WBS decomposition of activity price calculation. Example: Construction engineer (quantity surveyor) calculates the final price of a pavement with future realization by subcontractors. His activities are: studying blueprints, sending blueprints to subcontractors, receiving bids with quantities, reading and checking bids and quantities, comparing bids, making decision about best subcontractor and sending this information to supervisor. All of the activities demonstrated in the example can be improved by using a cooperation system which would require participation of all parties (construction engineer, supervisor and subcontractors). Or most of these activities can be handled more efficiently with computer automation which would require participation only from construction engineer and his company. 5. Abilities of computer to replace human work power As said in section 4, some activities are worth the effort towards automation. This paper presents an option for automation in a construction process of documentation flow. The design of this software and presumably also future trends in software development are based on finding a computer equivalent to WBS activity decomposition as we have seen presented in the previous part. Possible advantages of such computer equivalents are many. Let us bring some of them to attention. Information related activities (getting, receiving and searching for information) can be managed much faster and more precisely by information technologies (Weizenbaum, 1976). Reading of documents in written form is performed by OCR (Optical Character Recognition) and optical pattern recognition (Govindan, 1990). Decision making and problem solving done by computers can be achieved in situations with predefined boundaries (Greiff et al., 2013). Machine learning is a fast developing research area which focuses on the automated equivalent to learning (Hastie, Tibshirani, & Friedman, 2009). In foreseeable future, this research may bring results which simplify automated problem solving. 6. Economic aspects of computer automation in construction industry According to previous calculations, there is a 20% span of engineer s time suitable for work automation. It is difficult to state specific numbers concerning the number of construction engineers who are currently employed in construction industry. However, a rough number can be estimated using general statistics of the Czech Republic. There are approximately 433 thousand workers employed in construction industry in the Czech Republic i and the percentage of employees with university education is about 15%. Following from these numbers, there are approximately 64 thousand of construction engineers. Construction engineer earns 29000 CZK (Czech crown) per month on average. This salary represents approximately 39 000 CZK ii in costs for the employer (including taxes and insurance). To make the calculation more approachable, please note that there are 10.5 millions of inhabitants in the Czech Republic. i Source: http://www.czso.cz/csu/2012edicniplan.nsf/p/3115-12 ii Source: private database of salary (http://www.platy.cz/) 171

Using simple mathematics, automation of routine activities in construction industry represents a window of opportunity of direct costs with value of 570 million Czech crowns (21 million Euros) per year and million inhabitants. 7. Methodology software design Case study shows one way of efficient automation in construction process. It is focused on tender estimate comparisons in bids evaluation process. Such process is usually done by construction administrative engineer and it can be decomposed into several partial activities - elements: asking for bids, receiving bids from contractors (written form, pdf etc.), checking validity and calculation of individual bids, checking bid prices and bills of quantities with usual prices, comparing bids, decision making between possible contractors, announcing decision. There is a possibility to create computer equivalent for all these activities. All these equivalent parts will create a full application software. In figure 1, there is a diagram explaining document flow through the process described above. Presented algorithm is designed to process tables with a large amount of information like bills of quantities. Bills of quantities are evaluated against database of knowledge which is located at a publicly accessible place (server). Individual documents also enrich the database. Figure 1. Application for documentation automation design. A Receiving bid documents (manually or electronically). B Documents are scanned or loaded from memory and decomposed by optical pattern recognition into individual pieces of information (Gatos, Danatsas, Pratikakis, & Perantonis, 2005). C Original documents are decomposed into cells with relevant information in this case, item from bill of quantities: material, unit, number of units, price per unit and price total. D Individual pieces of information with coordinates of position in original document are transferred via OCR engine into electronic form. E Application detects what the purpose of information is by analyzing its position coordinates and type of information pieces F Every piece of information is tested for arithmetical and factual correctness (depends on numerical or text input) G After successful checking process individual pieces of information (in this case items from a bill of quantities) are compared with database of previous knowledge. H After successful checking process, information enriches database for future purposes. I Items are compared with database which implants the information into statistical context (prices, quantities etc.) based on exact match or keywords system (Agrawal, Chaudhuri, & Das, 2002). J All information is gathered and supplied into decision making process which can be based on a decision tree (Cha & Tappert, 2009). K If an item fails checking process it is pushed into manual sorting. L A report is created of decisions made based on documents from point A. M Database of previous knowledge is prepared for storage at a distant server for higher availability. N Database is enriched by processed information. Information contains not only prices and quantities, but also geographical location of the user etc. Database creates statistical samples of information (e.g. average prices, statistical distributions etc.). Advantages of described system against solely cooperative systems: Database of uses processed information to enrich itself, keeping itself up to date at all times. System is centralized. System can be used without the necessity for participation of all parties in construction process. With proper settings, the system can work without human interaction and therefore save time and costs. 172

8. Case study Presented algorithm was tested with good result. Test subject were bills of quantities for a family house with circa 200 items. Documents were delivered in paper form, scanned and saved in one folder. The algorithm detected tables in documents and decomposed them into individual cells. Optical pattern and character recognition engine (in this case engine Tesseract 3.03) transferred it into binary code with good results in 95% of items. Faulty items were detected in checking process and removed. Recognition and checking process took on average 22 seconds per page with circa 30 items (on Intel Atom 1,6Ghz laptop). Average time needed by a human worker to read and check the same number of items is 8 minutes and 40 seconds. Information database is created on SQL platform. When the algorithm was tested for comparing items with database and enriching database, it brought good results. However, database does not have a sufficient number of entries to create relevant results. In spite of this minor setback, case study shows that most parts of the design work and that it is viable. 9. Conclusion Let it be noted that there are at least two approaches to use of information technologies in construction industry cooperation and automation. Much attention is focused on cooperation (e.g. BIM, information systems etc.). However these systems have specific disadvantages. Mainly there is the need for a man to run them and they are only used with complex projects. To bring an element of contrast to the cooperation vs. automation debate, this paper presents the possibility of designing an automation application which can be used across the industry. An estimate is presented which strives to prove that automation is a way of improving work processes in 20% of engineers time. This time represents a window of opportunity with the approximate value of six billion Czech crowns in Czech construction industry. Such approach does not, however, suggest replacing a human worker with automation applications. Value of calculated direct cost at stake, could and should be redirected to increasing the quality of the final product, and towards improving work environment. Application design composed of IT alternatives to engineer s skills has been presented here. Application includes a self-learning algorithm with an information database which could be publicly accessible, and could provide relevant information for anyone. The main goal of the mentioned algorithm is processing documents with high content of information. Public institutions like universities etc. with proper funding could and should create similar applications and provide them to public on cloud services platform. Participant across industry would benefit from such accomplishment. Acknowledgement This work was supported by the Grant Agency of the Czech Technical University in Prague, grant No. SGS14/016/OHK1/1T/11. References Agrawal, S., Chaudhuri, S., & Das, G. (2002). DBXplorer: a system for keyword-based search over relational databases. Data Engineering, 5-16. Eastman, C. (2011). BIM handbook. New York: Wiley. Fewings, P. (2012). Construction project management: an integrated approach. New York: Spon Press. Gatos, B., Danatsas, D., Pratikakis, I., & Perantonis, S. (2005). Automatic Table Detection in Document Images. Pattern Recognition and Data Mining, 609-618. Govindan, V. K. (1990). Character recognition A review. Pattern Recognition, 671-683. Greiff, S., Wüstenberg, S., Holt, D. V., Goldhammer, F., & Funke, J. (2013). Computer-based assessment of Complex Problem Solving: concept, implementation, and application. Education Tech Research Development, 407-421. Harel, D., & Feldman, Y. (2004). Algorithmics: The Spirit of Computing. Boston: Addison-Wesley. Hastie, T., Tibshirani, R., & Friedman, J. (2009). The elements of statistical learning: data mining, inference, and prediction. New Your: Springer. Haugan, G. T. (2001). Effective Work Breakdown Structures. Management Concepts. Cha, S.-H., & Tappert, C. (2009). A Genetic Algorithm for Constructing Compact. Journal of Pattern Recognition Research, 1-13. McCabe, J., & Wissler, J. (2010). A time study of scientists engineers in the air vehicles directorate. Defense A R Journal. Parkhurst, H. B. (2011). Confusion, Lack of Consensus, and the Definition of Creativity as a Construct. The Journal of Creative Behavior, 1-22. Robinson, M. A. (2010). An Empirical Analysis of Engineers Information Behaviors. Journal of the American Society for Information Science and Technology, 640 658. Robinson, M. A. (2012). How design engineers spend their time: Job content and task satisfaction. Design Studies, 391 425. 173