Using Fuzzy Logic and the Management of Information To Increase the Performance of Construction Systems

Dean T. Kashiwagi

Del E. Webb School of Construction

Arizona State University

Tempe, Arizona

Introduction

There is a requirement in the construction industry and facility management arena to increase facility system performance and reduce the risk of premature replacement and maintenance and operational costs. Facility managers have traditionally used specifications and the "low bid" competitive award to attempt to minimize the cost of facility systems. The emphasis on cost has placed the construction industry in a dilemma of improving construction and facility system performance while reducing costs. The lack of a method to differentiate performance of constructors and their construction systems in a "low bid" environment has resulted in a decline of construction product performance. Facility owners, manufacturers and contractors of performing construction systems have identified a requirement for a new design, procurement, and construction delivery System which integrates performance information to deliver the "best available" performing contractors and systems for the most economical cost.

Declining Construction/Facility System Performance

The author (Kashiwagi:1991) modified a Hrebeniak and Joyce (1985) environmental determinism model into an "Industry Stability" model (Figure 1) and a "Low Bid Performance" model (Figure 2) to explain why the current industry structure and the competitive "low bid" award delivery system leads to a gradual trend toward minimal performance and lower performing construction products. "Industry stability" is defined as the ability of an industry to continually produce and improve a performing construction product (Kashiwagi:1994). Using the "Low bid" minimum specifications result in high competition and minimal standards (Quadrant 1 in Figure 1). Facility owners in both Quadrants 1 and 3 seek to move to the Performance Based Quadrant, either due to increased pressure to seek a higher level of performance or more competition. In both Quadrants 1 and 3, the perception is that performance is not being maximized. The author (1994) has defined "industry stability" as the relative size of the Performance Based Quadrant in relation to Quadrants 1 and 3. This is deducted due to the influence of the performing construction systems on the rest of the industry.

The first chart in Figure 2 shows four contractors with varying degrees of contractor/installed system performance. The top contractor is the highest performing contractor and #4 is the poorest performing contractor. The assumption made by the author is that contractor/system performance is related to investment and cost to the contractor. The minimum specification line is drawn between contractors #3 and #4. In this example, contractor #4 does not meet the minimum specification level of performance and cannot bid on the contract. As shown in the first chart, contractor #3 will be awarded the project. To become competitive, contractor # 1 drops the level of performance (and accompanying cost). In the second chart, contractor #1 is now the most competitive. Contractor #2 becomes more competitive in the third chart, dropping very close to the minimum performance requirements, thus securing the bid. Figure 2 shows the performance and cost of the contractors gravitating toward the minimum levels of performance set by the facility owner. This scenario leads to the following conclusions:

1.         The "low bid" award process assumes that all alternatives are equal in performance except for the cost.

2.         Facility owners define performance, and the contractors regulate performance to meet the requirements of the facility owners.

3          The "low bid" award delivery system does not give the constructors incentives for long range continuous improvement of product and performance.

The PBSRG used the above conclusions to shape a research program that performs the following:

1.         Define performance of construction systems.

2.         Collect performance information.

3.         Use the author's performance based procurement system (PBPS) to evaluate and procure construction  systems.

4.         Use performance information to assist manufacturers and contractors improve their performance.

5          Use performance information to assist facility owners minimize risk and cost.

Forward and Backward Chaining

Facility owners currently use "forward chaining" to meet facility system construction requirements. The facility owner or representative first analyzes the problem resulting in several possible solutions. An alternative solution is selected and specification created. A key point is that a specific facility owner and his representative may not consider all possible alternatives. The constructors who can deliver the specified systems then bid on the project, and the qualified low bidder (constructor L 1 in Figure 3) is awarded the contract. The drawback to this "forward chaining" process is that alternative #3.3 in Figure 3 was the best performing contractor/system.

The specification/low bid procurement process is a forward chaining solution process that does not result in high performance. Kashiwagi (1994) identified that a backward chaining solution process was much more efficient in producing high performing constructor/facility system performance (Figure 4). He stated that in situations where facility owners could easily define the requirement, they should use the management of information, fuzzy logic, and backward chaining to select the best performing contractor and installed facility system.

Performance Based Procurement Process (PBPS)

The author developed and tested the PBPS in 1991 and did further refinements and testing of the PBPS in 1993 1994. The author has collected performance information on industrial roofing systems for the past 12 years from 2,000 roofs in the United States and Europe. The first seven tests of the PBPS utilized roofing performance information. The PBPS process is shown in Figure 5. It includes the following steps:

1. The facility owner iden­tifies the requirement.

2. Constructors/designers propose alternative solutions, with accompanying costs and previous installations where they have implemented the proposed solution.

3. Performance data is collected on the alternative solutions.

4. A "fuzzy logic" relative distancing model is used to compare the alternatives and select the best constructor/facility system proposal.

The process can either be used as a pre-qualification or a one step competitive best bid delivery system. The system is inherently a continuous improvement system driven by information and thereby automatically pre-qualifies based on performance information.

“Fuzzy Thinking”

"Fuzzy thinking" is differentiated from traditional stochastic and parametric analysis by its simplistic method of "relativity." It uses the following tenants (Kosko:1994, Kashiwagi:1994):

1. Deterministic environment.

2. Every relationship or description is relative.

3. It is more efficient to describe a relative measurement of the relationship rather than to accurately quantify the actual relationship.

Research projects utilizing the PBPS show that unless the facility owner requirements and construction system data are dependent. This is reinforced by the concept of "relativity" in "fuzzy logic" and leads to the following conclusions:

1. Facility owner requirements cannot be predetermined.

2. Performance information is dependent with owner requirements.

The relative distancing model used is the "Displaced Ideal Model" (DIM) introduced by Zadeh (1982). The DIM is used instead of the Analytical Hierarchical Process (AHP) for the following reasons:

1          It is much more simplistic and easier to run.

2          It does not predetermine whether a weight scheme is correct.

3          It is less time consuming to run a large number of iterations.

4          It concentrates on prioritization of alternatives and identifying the performance requirement rather than the consistent weighting required for pairwise comparison in AHP.

5          DIM is more adept to assisting continuous improve­ment.

The PBPS is currently being run using off-the-shelf windowing software including a database, spreadsheet, and server interface. The software generates performance lines and allows the facility managers to weight the performance criteria to generate facility system requirements (Figure 6). Performance criteria include "combined performance criteria" which represent performance to facility managers. For example, one criteria represents the capability of a roofing contractor to install large roofs that do not leak, that satisfy the customer, and perform for an extended period of time. The repeated identification of roofing performance produces:

1.         An environment that can optimize contractor operation and material selection.

2.         Identification of roofing performance.

3.         An artificially intelligent system which can reproduce the decision making process of experienced facility managers.

Performance Information

As defined previously, performance information is defined by the facility owner. The following criteria is used on roofing requirements:

1. Proven performance period.

2. First cost and equivalent uniform annual costs.

3. Percent of roofs not leaking.

4. Percent of roofs not requiring maintenance.

5. Customer satisfaction.

6. Amount of traffic on a roof.

7. The amount of disturbance to the facility operations caused by the roof installation.

The facility owner's requirement is composed of a weighted scheme that prioritizes the performance criteria. This scheme is dependent with the actual performance information, requiring an iterative process to arrive at the final weighted scheme. By its inherent structure, the PBPS fulfills the requirements of construction "industry stability" which include:

1. Setting up entry/exit barriers.

2. Allowing the constructor to receive a fair profit.

3. Does not tolerate poor performance.

4. Encourages the constructors to do continuous improve­ment.

5. Allows total competition with no pre-qualification.

6. Gives the facility owner buyer protection.

Testing, Refining, Results

The PBPS has run seven times in 1994. Roofing constructors and installed roofing systems were procured six times. The seventh running procured a janitorial service. The roofing procurements were in Phoenix, Arizona (3), Tucson, Arizona (1), Indianapolis, IN (1), and Arlington Heights, IL, (1). The janitorial service was procured in Arlington Heights, IL. The PBPS was run for Motorola, Indianapolis Community Hospitals, and Morrison and Knudson (facility manager for IBM and the University of Arizona). The PBPS was also run for a facility manager of a large electronics manufacturing company in the Phoenix, metropolitan area who desired to remain anonymous.

The major conclusions from the roofing procurements were

1.         In five out of the six procurements, the facility managers selected roofing contractors and roof systems that they did not previously know about.

2.         In one of the PBPS, the facility manager selected a roof system type (single ply) which they had poor previous performance.

3.         In five PBPS’s, the facility managers procured the best available performing roofing system for a lower price than if they had used a specification and taken the low bidding constructor. The savings ranged from x.30/SF (30,000) to X2.00/SF 0200,000), considering only first cost.

4.         In four out of the six cases, the low bidder was not selected.

5.         Collection of the performance information can be accomplished in two weeks for 300 data points.

6.         The facility managers were all very comfortable with the selection of the contractors using the PBPS and performance information. Different facility managers have different facility requirements and the "best available" roof system is defined by the requirements of the facility owner.

7.         A couple of the facility managers were so impressed with the identified best performer that they have intentions to use the contractor for previous work without rerunning the PBPS.

8.         Manufacturers and contractors are currently doing continuous improvement to increase their system performance and competitiveness.

Conclusions/Recommendations

The Performance Based Procurement Process is a fully automated, information managing, continuous improvement tool which will select the best performing contractors which results in construction industry stabilization and the optimization of facility management funding. The PBPS cuts costs, increases performance, forces continuous improvement, and uses the latest techniques of the management of information, backward chaining, "fizzy logic," and artificial intelligence. The PBPS also redefines construction industry participants roles to maximize performance. It is easier to use, fully competitive, and reduces the time to procure.

The PBPS is being used by the Performance Based Studies Research Group at the Del E. Webb School of Construction at Arizona State University to also identify performances of general contractors, Job Order Contractors, mechanical contractors, and facility managers. The author has approached and discussed the use of Performance Based Procurement with facility managers from Honneywell, GM, and other major manufacturing groups to test the system in 1995. The PBPS will be run a minimum of six times in 1995. Another goal is to run the system out in a government procurement test. The author has applied for a patent on the PBPS and the computer application.

References

1.          Hrebiniak, L.G. and W.F. Joyce, (1985), "Organizational Adaptation: Strategic Choice and Environmental Determinism." Administrative Science Quarterly, 30, 336-349.

2.         Kashiwagi, Dean T., (1991), "Development of a Performance Based Design/Procurement System for Nonstructural Facility Systems." Arizona State University, Tempe, AZ.

3.         Kashiwagi, Dean T., (1994). Unpublished notes from "Performance Based Evaluation/Facility Management" Continuing Education course, Performance Based Studies Research Group, Del E. Webb School of Construction, College of Engineering, Arizona State University, Tempe, AZ.

4.         Kashiwagi, Dean T., Nuno, J. Pablo, and Moor, William, C, "Optimizing Facility Maintenance Using Fuzzy Logic and the Management of Information,16th International Conference on Computers and Industrial Engineering, March 7-9, 1994, Ashikaga, Japan, 404407.

5.         Kashiwagi, Dean T., and Ryan, Scott, (1994), Unpublished report "Stability of the Arizona Construction Industry."

6.         Kosko, Bart (1993), "Fuzzy Thinking, The New Sci­ence of Fuzzy Logic." Hyperion, New York, New York.

7.         Zeleny, M., (1982), Multiple Criteria Decision Mak­ing. New York: McGraw Hill.

Go to the  Home page for:

Associated Schools of Construction Proceedings of the Annual Conference.  Copyright 2003
For problems or questions regarding this web contact Tulio Sulbaran, Proceedings Editor/Publisher.
Last updated: September 09, 2004.