You are on page 1of 8

ISSN 2521-3806 ICBBD 2017

Vol.1 2-4 August 2017, Bangkok, Thailand

Practical Framework for Estimating and Planning Engineered Material


Fabrication: Case of Piping Spools

Ming Lu*, Meimanat Soleimanifar, Arash Mohsenijam

University of Alberta, Edmonton, Canada

* Corresponding author email: mlu6@ualberta.ca

Abstract
Over the past decade, the construction industry has been shifting to the paradigm of modular construction and
offsite prefabrication with high expectations on improving productivity and enhancing performances in
delivering projects. Mobile computing and big data technologies (such as Internet-of-Things, sensor networks,
and tablets) have also found significant applications in tracking assets and materials in the construction field. In
contrast, the long-standing mismatch in level of granularity between detailed cost estimating and rough cost
control is not yet resolved on most of construction projects as of date; the actual cost data rarely finds its way
back to inform the estimating of the next critical job. Practitioners still rely on the rule of thumb and “gut-
feelings” to make critical decisions, instead of data-driven, quantitative analytics. This study proposes a
framework to enhance the estimating and planning of construction projects featuring engineered material
fabrication and elaborates the framework using piping spool cases.

Keyword: Industrial Construction, Cost Estimation, Actual Labour Cost , Project Productivity

1. Introduction
Modern construction projects have been shifting to modular and off-site prefabrication technologies for better
quality and productivity benefits (Eickmann and Fagerlund 1999). A mismatch in granularity between
estimating/planning systems and job costing systems is commonplace in the current practice of construction
management. This mismatch is rooted in the fundamental discrepancy between cost estimating and cost control
in regards to the level of granularity of systems and data (Peurifoy and Oberlender 2002). Job costing data is
generally broad stroked while estimating and planning deal with much detailed information. Productivity data
are not accurately tracked in the real world at a level sufficient for desired productivity modeling analysis (Song
et al. 2004). Such a mismatch undermines the value of the precious historical data representing company’s
actual performance, thereby preventing the formation of a feedback loop, which takes full advantage of recorded
data for decision support in informing estimating and planning on new projects and improving cost advantages
and strategic competitiveness of a construction company.

The current practice of estimating still relies on the “benchmark” data which at best denotes the average
productivity performance in the industry or company. Such “benchmark” data are often deemed too
conservative and are commonly overridden by experience and gut feeling in consideration of perceptions on
company-specific actual performances and productivity situations (Van Vilet 2011). The overall process
remains subjective and uncertain, with estimates missing or hitting on actual costs being much analogous to the
event of “making bets in a casino”.

On the other hand, applying new sensor technology to track detailed job operations in construction is still
prohibitively expensive in terms of technology investment and overhead cost such that scaling up the technology
application for implementation is rare in practice. Despite the decrease in the cost of implementing
technological devices [such as RFID or wireless sensor networks (WSN)], solutions remain expensive and
demand specialists to design and calibrate; For instance, deploying and calibrating radio frequency based
position systems for a fabrication shop or a construction site (Soleimanifar 2016). Crews need to be trained and

41
ISSN 2521-3806 ICBBD 2017
Vol.1 2-4 August 2017, Bangkok, Thailand

stay committed to tagging, detagging, and scanning pieces and assemblies in tracking material flowing and
conversion processes during their work. In addition, tracking laborers on the shop floor or in the construction
site is not generally accepted practice due largely to ethical issues of privacy infringement.

The other challenge lies in the fact that a detailed work plan is a prerequisite for implementing the advanced
technologies. The workface plan depicts pieces’ flows and conversions in a fabrication shop or in a construction
site, facilitated by labor or equipment resources. The analytical methodologies have advanced significantly in
research. However, the absence of the workface plan automatically generated by computer in a scientific fashion
is partly attributed to missing detailed operations data. In practice, the workface plan is commonly made by
operations personnel (i.e. superintendent, foreman, and tradespersons) by “sketching on napkins.” The process
of making and communicating a workface plan needs to be flexible and straightforward in order to cope with
constant changes and overwhelming complexities of the real world, which remains more of art than of science.
Nonetheless, many opportunities for work improvement through applying quantitative optimization analyses
could have been missed. In short, the art of workface planning stays with the master planner him/herself, while
transferring the knowledge gained and lessons learned from previous projects to inform the bidding and
execution of new projects remains difficult. Passing know-how and experience from masters to novices relies on
the ageless apprentice training model, with no data-driven, analytics-based decision support.

2. Proposing a solution framework


A reliable source of productivity data already exists in most construction companies in terms of project
attributes, progress measurements, and labor costs. Historical data serve as a basis for productivity models that
can be utilized to predict productivity of future projects (Song and Abourizk 2008). On previously completed
projects, historical data on design details (drawings or models) along with material takeoff data
(pieces/components) are generally available during the course of estimating the labor cost (man-hours). Actual
labor expenditures are generally organized by a comprehensive list of work items performed to handle and
connect pieces, turning them into assemblies ready for shipment. The level of detail in material cost and quantity
should match with the information available in estimating stage of project, which are obtained either by manual
takeoffs of drawings or by automated takeoffs in BIM (Mohsenijam and Lu 2016).

The list of work items represents the granularity of work breakdown acceptable to estimators in the current
practice. Materials quantity takeoff is performed based on design details, followed by converting materials'
quantity to work content in man-hours, given a labor productivity rate (man-hour per unit) is known for each
work item.

In the current management practice, the actual cost data are collected as a total for a particularly defined “Cost
Center”. A cost center can be a division, a grouping of the same type of work, or a project for tracking job cost
and client payment; the data are in general taken as actual cost of completed projects, representing the cost and
productivity performances of the company. Many companies capture labor productivity information only at the
level sufficient to cater to the requirements of the project owners (Mohsenijam and Lu 2016), which usually
does not align with the level of details required in estimating. In order to resolve the granularity mismatch
between estimating and job costing, an analytical framework is proposed to prorate actual total man-hours
recorded in one “cost center” over relevant work items that are performed in materializing all the construction
deliverables associated with the “cost center”.

∑ ���� � �� � �� (1)

Given Qip is the material quantity takeoff for work item i on cost center p; Ui is the labor productivity rate on
work item i (MH/Unit) –which is the productivity ratio to convert material (in Unit) takeoff into work content
(in man-hours); Wp is the actual total man-hours recorded on the current “cost center” p.

The objective of analysis is to fix Ui as constant (the optimum value) which would result in the least error
between estimated values of ��� and actual values of Wp; given the presence of noise in practical data (error or
missing data) the “least error” can still be relatively large; thus, a statistical characterization on the error is
required so as to represent the uncertainty in the modeled output ���.

42
ISSN 2521-3806 ICBBD 2017
Vol.1 2-4 August 2017, Bangkok, Thailand

An alternative perspective is to fix the likely range of Ui as a distribution range while ܹ‫݌‬Ԣ = Wp is set as a
constraint in the least square optimization of regression analysis on Eq. (1). Note, depending on the quality and
quantity of available historical data, the range estimated for Ui can be a relatively wide distribution; thus, the
confidence interval or percentiles for the distribution lends decision support to the estimator in fixing a most
likely point estimate on the range when bidding for a new project.

3. The list of work items is the foundation


A comprehensive list of commonly applied work items involving diverse trades of laborers in handling pieces
and connecting pieces needs to be identified based on the current estimating and planning practice in
collaboration with subject experts. Once Ui is fixed for each work item, being a constant or a distribution, the
estimator will be guided by actual performance data in estimating and planning new projects. Estimating Wp
uses the same Equation (1) on a given new project. The estimator first takes off the material quantity in
respective units of measure for all the relevant work items in fabrication of the new preassembly. Next, by
factoring the productivity situation and job complexity on the new project, the estimator presents a range
estimate on Wp with statistical confidence or selects Ui from the range with statistical confidence. Finally, the
work content in man-hours (Wp) is determined for the new project.

4. Collecting detailed operations data: resource use and time


Once the work content in man-hours is fixed on each work item involved in a work package (activity) of a
fabrication or construction project, how to break up the value of man-hours into the number of workers and the
time duration of the work package? Addressing this question is equivalent to collecting detailed operations data
on the shop floor or in the construction site. It is not as straightforward to break up by applying the mathematical
relationship of MH = M x H.

The proposed framework will rely on the use of practitioners’ experience to map out work flows in activity-on-
node network diagrams on a particular project (which depicts various materials or laborers flowing from work
item to work item by changing time and location states in the construction process). The AON network model
simulates the execution of work items by a certain crew (crew size known) on related work packages. The
project completion time (hr) multiplied by the size of the crew deployed gives the total supply of crew resource
in man-hours, which is related to the sum of Wp on all the work items, while on each work package, the activity
time multiplied by the number of workers involved is constrained to equate with the estimated value (Qip x Ui).

It is worth mentioning the AON based work flow simulation is capable of showing transparent logic and
processes. AON shows the sequence of handling and connecting activities, which are compatible with the major
work items identified in estimating and planning. The work breakdown needs to be confirmed by experienced
practitioners. The outcome will be a calibrated AON model for each project, which is essentially a sufficient
simulation of the project, consisting of the work breakdown, work items, and the logic in articulating them. The
work content (man-hours) in each work item and in the total project closely match with the crew resources
provided to each corresponding work package and to the whole project. As such, the time event scheduling of
construction activities would produce a feasible work plan for the crew to execute the project, subject to
allocated crew resource capacity (shop labor resource supply) and work content requirement on each work
package (project labor resource demand).

In short, such AON network models calibrated to previous projects and historical data effectively guide work
breakdown, material quantity takeoff and work hours takeoff in estimating a similar new project in the future.
More importantly, those calibrated AON network models can be utilized as templates to facilitate workface
planning and project scheduling on a new project; they will potentially allow for project-production integrated
optimization analysis aimed to balance resource supply and demand in a given production facility through
unifying project scheduling and production planning.

43
ISSN 2521-3806 ICBBD 2017
Vol.1 2-4 August 2017, Bangkok, Thailand

5. Case Study
The selected spool fabrication project is part of an oil sands upgrading facility expansion in Fort McMurray,
Alberta, Canada. All the spools were pre-fabricated in one fabrication shop located in Edmonton, Alberta. After
the spools were finished, they are transported to the module yard for further assembly. In this case study, three
sample spools from the project are selected to build a test case, so to verify the feasibility of the proposed
methodology for determining the productivity ratio for the spool fabrication operations (i.e., mainly fitting ,
welding and handling). The isometric drawings for the three spools are shown in Figure 1. To illustrate the
problem, Table 1 gives the material takeoff of pipe spools depicted in Figure 1, showing the required materials,
including material types and quantities. Note, pipe size is specified with two non-dimensional numbers, namely:
a nominal pipe size (NPS) and Schedule (Sch.).

Figure 1. Pipe spool fabrication drawings

NPS is a North American standard for representing the pipe diameter based on inches. “Sch.” numbering is a
North American standard that denotes the wall thickness of a pipe or pipe fitting, such as Sch. 10, Sch. 40, Sch.
60, Sch. 80, standard wall (STD), extra strong (XS) and double extra strong (XXS). Higher schedules mean
thicker walls of pipe for withstanding higher pressures.

Table 1. Required materials for fabricating the three sample spools


Quantity
Material Name
Spool 1 Spool 2 Spool 3
6'', Sch.40 Pipe 12.1 (m) 5.4 (m) 7.3 (m)
2'', Sch.80 Pipe 1.1 (m)
3'', Sch.40 Pipe 1.4(m)
2'', Sch.80 Elbow 1
6'', Sch.40 Elbow 2 2
2'', Sch.80 Flange 1
3'', Sch.40 Flange 1
2''×6'', Sch.80/Sch.40 Olet 1
6'' Hydro pipe 1
6''×3'', Sch.40 Reducer 1
6'', Sch.40 Tee 1

The activities required for fabricating Spool 1 are tabulated in Table 2. The same way, fabrication activities for
Spool 2 and Spool 3 can be defined respectively. The labor requirements of each activity are listed in Table 2 as
well. For illustration purposes, Figure 2 implies the fabrication sequence for pipe spool 1.

44
ISSN 2521-3806 ICBBD 2017
Vol.1 2-4 August 2017, Bangkok, Thailand

Table 2. fabrication activities for Spool 1


Activity ID Activity Name Labor
A1 Handling 2’’ elbow -
B1 Handling 2’’ flange -
C1 Fit and weld 2’’ elbow and 2’’ flange (sub1_1) 2 Fitters and 1 Welder
D1 Handling sub1_1 1 Handler
E1 Handling 2’’ pipe 1 Handler
F1 Fit and weld sub1_1 with 2’’ pipe (sub1_2) 2 Fitters and Welders
H1 Handling 6’’ pipe 2 Handler
I1 Handling 6’’ pipe 2 Handler
J1 Fit and weld two 6’’ pipes together 2 Fitters and 1 Welder
G1 Handling sub1_2 1 Handler
K1 Handling the olet -
L1 Fit and weld sub1_2 with olet (sub1_3) 2 Fitters and 1 Welder
M1 Handling 6’’ pipe 2 Handlers
N1 Handling sub1_3 1 Handler
O1 Fit and weld sub1_3 with 6’’ pipe 2 Fitters and 1 Welder

Figure 2. Pipe spool 1 activity breakdown

The logical relationships (i.e., technological constraints) between activities in connection with fabricating a
particular spool are represented in the AON network shown in Figure 3, which closely mirrors the actual work
sequences executed in the fabrication shop of the partner company. Table 3, 4 and 5 demonstrate the cost
estimate summary for welding each pipe spool using industry benchmark man-hour rates (Page and Nation
1976.)

45
ISSN 2521-3806 ICBBD 2017
Vol.1 2-4 August 2017, Bangkok, Thailand

A1
C1 D1

B1 F1 G1

E1 L1 N1

K1
N1

H1
J1 M1

I1

Figure 3. AON network for fabrication spool 1

Table 3. Cost estimate summary for pipe spool No.1

Item Size Wall Thickness Quantity MH/Per Unit of Total Man-Hours


measure
Weld-neck flange 2" Sch 80 1 1.05 1.05
Olet Weld 2" Sch 80 1 4.20 4.20
Butt Weld 6" Sch 40 1 1.80 1.80
Butt Weld 2" Sch 80 2 0.90 1.80
Total Man-hour for the spool 8.85

Table 4. Cost estimate summary for pipe spool No.2

MH/Per Unit of
Item Size Wall Thickness Quantity Total Man-Hours
measure
Weld-neck flange 3" Sch 40 1 1.70 1.70
Butt Weld 3" Sch 40 1 1.10 1.10
Butt Weld 6" Sch 40 6 1.80 10.80
Hydro Weld 6" Sch 40 1 1.80 1.80
Total Man-hour for the spool 15.40

Table 5. Cost estimate for pipe spool No.8

MH/Per Unit of
Item Size Wall Thickness Quantity Total Man-Hours
measure
Butt Weld 6" Sch 40 3 1.80 5.40
Total Man-hour for the spool 5.40

Table 6 contrasts the total actual MH for each pipe spool derived for companies’ existing job costing system and
the total estimated MH from Page and Nation (1976). The average adjustment factor is 0.43. Using this factor,
the total actual MH is prorated to all work items in each pipe spool based on Page and Nation man-hour rates.

46
ISSN 2521-3806 ICBBD 2017
Vol.1 2-4 August 2017, Bangkok, Thailand

Table 6. fabrication activities for Spool 1


Pipe Spool No. Actual MH Page and Nation Adjustment Factors
1 3.00 8.85 0.34
2 8.58 15.40 0.56
3 2.17 5.40 0.40

Table 7 shows the proration process for pipe spool No. 1.

Table 7. The proration process for pipe spool No. 1


Wall MH/Per Unit of Adjustment Total Man-
Item Size Quantity
Thickness measure Factor Hours
Weld-neck flange 2" Sch 80 1 1.05 0.43 0.45
Olet Weld 2" Sch 80 1 4.20 0.43 1.81
Butt Weld 6" Sch 40 1 1.80 0.43 0.77
Butt Weld 2" Sch 80 2 0.90 0.43 0.77
Total Man-hour for the spool 3.81

Similarly, the proration process is performed for the other two spools. Table 8 shows the actual MH and
Prorated MH. The average error is 0.97 MH. The results indicate that the proposed solution framework is able to
accurately determine the labor productivity rate on work items using the actual MH from the existing costing
system of the company.

Table 8. fabrication activities for Spool 1


Pipe Spool No. Actual MH Prorated MH Error MH
1 3.00 3.81 0.81
2 8.58 6.62 1.96
3 2.17 2.32 0.16

With the labor requirements of each activity in Table 1 and prorated man-hours for each work item, the planner
would be able to determine the duration for completing each work item in AON.

6. Conclusion
Mega industrial projects frequently suffer from cost overruns and schedule slippages. As projects increase in
size, and become more complex, managing these projects becomes more challenging. Therefore, in
modularization of heavy industrial construction, the availability of actual labour cost and productivity data is
important to accurate pricing and profitable billing but also provides a valid basis for cost estimating and
progress monitoring on similar projects and also for shop production scheduling.

This research study taps into the existing knowledge of industrial construction projects while also looking into
the existing practice of industrial construction cost estimation for pipe spool fabrication processes. A solution
framework is proposed for determining labour productivity norms for work items involved in pipe spool
fabrication using actual labour hours recorded in the exsisting company’s costing system and the benchmark
labor hour norms from Page and Nation (1976).

The proposed system is not only able to provide labour hours and time duration for spool fabrication work
items, it also provides estimators with calibrated labour productivity norms based on company’s actual
performance on historical projects.

47
ISSN 2521-3806 ICBBD 2017
Vol.1 2-4 August 2017, Bangkok, Thailand

REFERENCES
Eickmann, J. A., and Fagerlund, W. R. (1999). "Prefabrication and preassembly trends and effects on the
construction workforce." University of Texas at Austin.
Peurifoy, R. L, and Oberlender, G. D. (2002). Estimating construction costs. 5th ed. Boston: McGraw-Hill.
Van Vliet, M. S. J., (2011). DACE Labour Productivity Norms - The New “Gulf Coast?”, AACE International
Transactions
Soleimanifar, M. (2016). “Integrated Project Management Framework for a Pipe Spool Fabrication Shop” . PhD
dissertation, Canada
Song, L., and Abourizk, S. M. (2008). “Measuring and Modeling Labor Productivity Using Historical Data.”
Journal of Construction Engineering and Management, 134(October), 786–794.
Mohsenijam, A., and Lu, M. (2016). “Achieving Sustainable Structural Steel Design by Estimating Fabrication
Labor Cost Based on BIM Data.” Procedia Engineering, 145, 654–661.
Song, L., Allouche, M., and AbouRizk, S. (2004). “Measuring and Estimating Steel Drafting Productivity.”
Construction Research Congress 2004.
Page, J., Nation, J. (1976). “Estimator‘s Piping Man Hour Manual”, Gulf publishing company, USA

48

You might also like