1. 1 7. R EPETITIVE C ONSTRUCTION Objective: To understand how production and production rates are affected by repetition of tasks, and to learn how to plan for tasks that are highly repetitive. Summary: 7.1 Introduction 7.2 Linear Scheduling 7.3 Learning and Forgetting Effects

2. 2 7.1 I NTRODUCTION Many tasks in construction are repetitive, and these can be divided into two basic types: –Discrete Repetition (where discrete units of an item are built many times over): building many similar housing units; building multiple (similar) floors on a high rise building; –Continuous Repetition (where the item being constructed is one continuous unit): laying a long pipeline; constructing a tunnel; building a long RC retaining wall; –Note, tasks that are continuously repetitive are often performed in discrete units: an RC retaining wall may be built in sections; while a tunnel boring machine will operate continuously, lining the tunnel behind it may be done in discrete sections.

3. 3 Fig. 7-1: Ladder Representation of Highly Repetitive Tasks 7.2 L INEAR S CHEDULING CPM is not an appropriate planning tool for tasks that are highly repetitive: –activity networks can become complicated; –although activity networks can be simplified with a ladder, if there is a lot of repetition, then: it is difficult to see where phased tasks may interfere with each other the internal critical path could become complicated if the time to repeat a task varies. Each activity is made up of lots of sub-activities. We do not know where the interferences (and thus the critical path) will occur internally on a ladder.

4. 4 Linear scheduling provides a very simple yet very effective means of modeling highly repetitive tasks: Fig. 7-2: Simple Velocity Diagram for a Pipeline Project TIME WORK COMPLETED LAYOUT EXCAVATELAY PIPE BACK FILL Time Buffer (Free Float) between 2 activities Space buffer between 2 activities BACK FILL started late to avoid interference with LAY PIPE and to provide a minimum space buffer at their point of convergence LAY PIPE started late to provide a space buffer between itself and EXCAVATE Minimum space buffer between LAY PIPE and BACK FILL Gradient of line = production rate. Time Now read off stage of completion read off stage of completion The starts of both LAYOUT and EXCAVATE are critical and the whole of LAY PIPE

5. 5 Linear scheduling allows us to determine ways of reducing the project duration. This may involve: –adding more crews (including equipment) to certain activities; –adding more crew members (and/or other type of resources such as equipment) to certain activities; and –adding more efficient/productive resources to certain activities.

6. 6 How can the project duration be reduced? Fig. 7-3: Crashing (compressing) Project Duration TIME WORK COMPLETED LAYOUT EXCAVATELAY PIPE BACK FILL started earlier than before in an attempt to finish sooner Minimum space buffer between LAY PIPE and BACK FILL finishes no sooner Q: Can we start activity BACK FILL sooner? A: Yes, but will not reduce project duration, yet will increase costs as introduces idle time where it slows down to maintain space buffer with LAY PIPE.

7. 7 Can reduce project duration by adding extra crews: –if do this for one activity, it would have to be LAY PIPE Fig. 7-4: Adding Extra Crews to Reduce Project Duration TIME WORK COMPLETED LAYOUT EXCAVATE Minimum space buffer between LAY PIPE and BACK FILL LAY PIPE BACK FILL Second crew can start when there is enough space between (1) the two LAY PIPE crews; and (2) 2nd LAY PIPE crew and EXCAVATE crew. BACK FILL can start and finish earlier, reducing project duration.

8. 8 Can reduce project duration by adding extra crews to adjacent tasks: Fig. 7-5: Adding Extra Crews to Two Adjacent Tasks TIME WORK COMPLETED LAYOUT LAY PIPE BACK FILL BACK FILL can start and finish earlier, reducing project duration. EXCAVATE Minimum space buffer between LAY PIPE and BACK FILL

9. 9 Alternatively, sometimes, can reduce project duration by adding extra members to a crew: Fig. 7-6: Adding Extra Crew Members to Reduce Project Duration TIME WORK COMPLETED LAYOUT EXCAVATE LAY PIPE BACK FILL BACK FILL can start and finish earlier, reducing project duration. More crew members can increase production rate

10. 10 Finally, how can project duration be reduced without adding more resources such as crews, equipment, or crew members? Fig. 7-7a: Reducing Project Duration TIME WORK COMPLETED LAYOUT EXCAVATELAY PIPE BACK FILL In this problem, we have an additional activity (COMPACT). Its minimum buffer with the preceding activity is at its start. COMPACT How can we start activity COMPACT sooner without adding more resources to any activity?

11. 11 One solution is to reduce the crew size for activity BACKFILL. Fig. 7-7b: Reducing Project Duration by Delaying a Crew TIME WORK COMPLETED LAYOUT EXCAVATELAY PIPE BACK FILL The reduced crew size on BACKFILL means its production rate will be reduced and thus it needs to start sooner to ensure that it finishes at its early start time. This, in turn, allows COMPACT to start sooner and thus finish sooner. COMPACT

12. 12 Another solution is to introduce a delay to the crew for activity BACKFILL. Fig. 7-7c: Reducing Project Duration by Delaying a Crew TIME WORK COMPLETED LAYOUT EXCAVATELAY PIPE COMPACT The activity BACKFILL start sooner but is then interrupted to prevent interference with LAY PIPE This, in turn, allows COMPACT to start sooner and thus finish sooner. BACK FILL

13. 13 Not all tasks progress at a constant rate: therefore, need to allow for this in planning task schedules. Fig. 7-8: Production Curves may be Non-Linear TIME WORK COMPLETED LAYOUT BACK FILL LAY PIPE EXCAVATE The nature of the work may change over the project, for example, the depth of the excavation may vary over the length of the pipe run. Learning effects can cause significant acceleration on some tasks

14. 14 Identifying the critical path is straight forward: –it must go from the start to the end by some route Fig. 7-9: Determining the Critical Path(s) LAYOUT TIME WORK COMPLETED LAY PIPE BACK FILL EXCAVATE CRITICAL PATH Easiest to determine this working backwards from the end of the project. The critical path takes this route because the gaps are equal to the minimum value for the space buffers. ARTIFICIAL CRITICAL PATH This represents the portion of a task that is critical because it is artificially made to start at its late start time (according to the logic of the activity dependencies, it could start sooner).

15. 15 There could be some tasks that are performed at a fixed location: Fig. 7-10: Production Curves may be Non-Linear TIME Distance LAYOUT BACK FILL LAY PIPE EXCAVATE Second pole is critical as it has forced EXCAVATE (which is critical) to be delayed Remove two utility poles

16. 16 7.3 L EARNING AND F ORGETTING E FFECTS Research shows that whenever a task is repeated many times in construction, there is usually a corresponding reduction in how long it takes to perform: –this is referred to as the learning (or experience) effect; –if the task is repeated with minor interruptions: the curve can be approximated by a simple mathematical function (a negative logarithmic curve); –whenever there are significant interruptions: some of the learning is lost and performance regresses part of the way back up the curve; this is referred to as the forgetting effect.

17. 17 Fig. 7-11: Learning and Forgetting Curve Repetition Number Time to Complete a Task 123456789101213141516171819202122 A delay between the 12th and 13th repetition of the task led to forgetting. The new learning curve has the same form as an earlier part of the original learning curve - in effect learning has regressed. In this example, the 13th repetition has regressed back as if it were just the 4th repetition.

18. 18 There are many implications for the learning and forgetting effects: –you can plan for work to accelerate if it is repetitive: thus, fewer crews may be required to get the work completed on time; –you should not swap crew members, and you should avoid changing crews, since the benefits of learning will be lost: the learning curve will remain flat in the worst case; –if work is falling behind schedule, you should add crews (or crew members) early so that they have more time to learn: this way, fewer crews will be needed and will operate at a higher productivity (so that costs come down).

19. 19 Theory: –the learning curve can be approximated by a negative logarithmic function: for every doubling in the number of repetitions, there is a constant decrement in CAT (the Cumulative Average Time to complete those tasks): 1125.0 days125 days - 275.0 days100 days80% 456.7 days80 days80% 844.3.7 days64 days80% REPETITON TIME t TO COMPLETE TASK CAT DECREMENTAL CONSTANT IN CAT FORDOUBLING IN REPETITIONS

20. 20 thus, in theory, plotting the logarithm of CATn against the logarithm of n should produce a curve that approximates a straight line: Fig. 7-12: Log-Log Learning Curve Logarithm of n-th repetition Logarithm of CATn 0 Slope (s) Intercept (c)

21. 21 If we know s and c then we can predict: –the time, T, to complete n repetitions of a task; –the time, t, to complete the n-th task; and –the number of repetitions, n, we can complete in a given period of time, T; From this, we can then also determine how many crews, or extra crews are needed to complete the work on time. How do we find s and c ?: –from published data: see lecture 7 notes; –from our own site data: if we know how long it took to perform the first few repetitions of a task, then we can plot on a log-log graph and find a best fit line;

22. 22 Fig. 7-13: Finding Approximation of Log-Log Learning Curve. Logarithm of n-th repetition Logarithm of CATn 0 Best fit line through points. This can be achieved using linear regression. Knowing the best fit line, we can now determine appropriate values for s and c. = observed values