Work Zone Mobility and Safety Program

EA 08-472214

slide 1

Interstate shield with the words I15 Ontario and 'OntFix'.

Jonathan den Hartog, P.E.

View the PDF Version [1.8 MB]
Download the free Adobe Reader to view PDFs You will need the Adobe Reader to view the PDFs on this page.

slide notes

  • Introduce yourself.
  • $52 million project pavement rehabilitation project
  • Begun in April 2009. Estimated completion by the end of this year.

slide 2

Project Location

Map highlighting the I-15 project area in Ontario, CA as well as identifying local landmarks such as malls, a racetrack, an airport, and railroad tracks.

slide notes

  • Ontario is about 40 miles east of downtown LA
  • In Addition: Major route for Las Vegas traffic. No break on weekends.

slide 3

Project Location

Aerial photo of project area including a notation that the I-10, which intersects with I-15, is an8 lane highway. I-15 has an ADT greater than 200,000, 6% of which are trucks. SR60, a 6-8 lane major arterial, also intersects I-15.

slide notes

  • Again, no break on weekends

slide 4

Project Features

  • Median Paving
  • Bridge Widening
  • New Median Barrier
  • Shoulder/Ramp Rehab
  • Pavement Rehabilitation
    • 12 ln-mi lane replacement
    • Random slab replacements
    • Includes connectors
    • Precast Pavement (Super-Slab)

Photo of traffic on the I-15.

slide notes

  • Picture is non-rush hour, on a weekday. Saturday would be similar. All days can have traffic jams
  • Bridge widening was to close the gap between NB and SB bridge decks.

slide 5

Traffic Volumes

Line graph shows ADT on the I-15. Northbound weekday ADT is 117,000; northbound Saturday ADT is 107,000; southbound weekday ADT is 103,000; southbound Saturday ADT is 105,000.

slide notes


slide 6

Construction Sequence

Simple diagram indicates that there are four southbound lanes and four northbound lanes on the I-15. The construction roadbed on the northbound segment will involve erecting a temporary concrete varrier on the far right lane, and two additional lanes are positioned to the right of the barrier. The lane adjacent to the right-hand travel lane is labeled continuous lane reconstruction 2 and the lane to the far right is labeled continuous lane reconstruction 1.

slide notes


slide 7

Rapid Weekends

  • 5 Major Stages,
    • 25 sub-stages
  • 410 Working Days (~2 yrs)
  • 55-Hour Weekend Closures
    • Beginning late Friday evening
    • Ending early Monday morning
  • Approximately 30 weekends
  • ~8 full roadbed closures

slide notes

  • Needed CA4PRS to justify this approach.
  • Also, wanted to include incentive to reduce number of weekend closures. Needed to justify this.

slide 8

Typical Closure

Technical drawing of a roadway typical closure on the I-15.

slide notes


slide 9

CA4PRS Study

  • 2 Phase Study
    • Alternative Analysis And Comparison
    • Detailed Study of Preferred Alternative
  • Performed by consultant sub

Screenshot of the PCCP Deterministic OntFix Study project details input screen.

slide notes

  • The subcontractor was Dr. E.B. Lee
  • The CA4PRS study was done in conjunction with a traffic modeling study, performed by a different sub.
  • At the time, CA4PRS did not include the Work-Zone Analysis or Agency Cost modules. These efforts had to be done separately.
    • Traffic analysis was done by EB Lee (in addition to the traffic modeling study) using the same Demand Capacity Model now incorporated into the software.

slide 10

CA4PRS Study Purpose

  • Validate the alternative chosen
  • Provide a detailed estimate of:
    • Working days for the project
    • Number of closures needed for each stage
  • Basis for Incentive/Disincentive

$150,000 per saved closure

(Max $900,000)

$175,000 per extra closure

slide notes

  • $150,000 is the average delay damages calculated.
  • The disincentive includes $25,000 for the public awareness campaign to alert the public about the closure.
  • Incentive is applied for every closure less than 27 used. Disincentive for every closure over 32.
  • Did not use the software for LCCA.

slide 11

Alternatives Analysis

Scenario Closure Duration Traffic RUC** ($M) Traffic Delay**(min) Agency Cost ($millions) TotalCost** ($millions) Cost Ratio
1. Original 35 weekends 3 16 78 79 100%
3. Contraflow 55-hr Weekend 35 weekends 119 363 83 123 156%
4. Progressive Continuous 8 weeks 123 363 77 118 149%
5. Traditional Nighttime 1,220 nights 133 22 88 133 168%
6. CSOL 55-hour weekend 20 weekends 69 363 60 83 105%
** Total Cost = 1/3 RUC + Agency Cost

slide notes

  • Go over columns. Delay is per closure.
  • 2nd alternative dropped out because of safety concerns.
  • Alternative 1 required bridge widening
  • Alternatives 3,4 would have serious traffic delay impacts, and would be difficult to determine peak dir., but might be tolerable if the duration was short.
  • Alternative 5 would take a long time, and still result in a inferior product.
  • Alternative 6 was close in value to the original, but would have higher delays, and was not preferred by maintenance.

slide 12

Road User Costs/Delay

Scenario Demand Reduction 1,500 vphpl* capacity
Queue (Miles)
1,500 vphpl* capacity
Delay (Minutes)
1,500 vphpl* capacity
RUC** $(Millions)
1,700 vphpl* capacity
Queue (Mile )
1,700 vphpl* capacity Delay (Minute ) 1,700 vphpl* capacity
RUC** $(Millions)
1 – Original 20% 8 61 20 2 18 2
1 – Original 30% 2 16 3 0 0 0
3 – 55-hour Weekend 30% 51 363 119 34 210 63
3 – 55-hour Weekend 40% 25 179 45 13 81 17
4 – Progressive Continuous 30% 51 363 123 34 210 51
4 – Progressive Continuous 40% 25 179 47 13 81 13
5 – 8-hour Nighttime 5% 8 57 418 - - -
5 – 8-hour Nighttime 10% 3 22 133 - - -
06/01 – CSOL (Weekend) 30% 51 363 69 34 210 36
06/01 – CSOL (Weekend) 40% 25 179 25 13 81 10
06/02 – 2CSOL (Nighttime) 5% 8 57 120 - - -
06/02 – 2CSOL (Nighttime) 10% 3 22 38 - - -
* vphpl: vehicle per hour per lane
** RUC: Road User Cost

slide notes

  • Note that this was done in a spreadsheet that formed the basis of the module in CA4PRS. It was not a part of CA4PRS when we did our analysis.
  • On at least one weekend, our vphpl approached 2000.
  • Actual delay was generally between 10-15 minutes during peak hours.

slide 13

Road User Costs/Delay

Graphic comparison of RUC and delay for original, 55-hour weekend, continous, nighttime, and CSOL scenarios.

slide notes


slide 14

Stage Analysis (Sample)

Stage Station Start Station End No. of Lanes Length
Length (lane-km) Rehab Type Total
55-hour Closures Estimate
4B, 4C 836+81 837+81 1 100 0.1 CLR 2.5 2
4B, 4C SB I-15 Conn WB SR-60 SB I-15 Conn WB SR-60 2 773.2 1.546 CLR 2.5 2
4B, 4C SB I-15 Conn EB SR-60 SB I-15 Conn EB SR-60 2 1430.5 0.858 RSR 2.5 2
2A 7+40 11+79 2 439 0.878 CLR 1.28 1
2A Jurupa On-ramp SB I-15 Jurupa On-ramp SB I-15 2 500 0.4 ACR 1.28 1
2B 20+90 22+77 2 187 0.374 CLR 3.01 3
2B 22+77 28+51 1 574 0.574 CLR 3.01 3
2B 28+51 33+03 2 452 0.904 CLR 3.01 3
2B 20+96 28+95 1 799 0.24 RSR 3.01 3
2B WB I-10 CONN SB I-15 WB I-10 CONN SB I-15 1 337.7 0.338 CLR 3.01 3
2B SB I-15 Jurupa Off-ramp SB I-15 Jurupa Off-ramp 2 500 0.4 ACR 3.01 3
2B WB I-10 Conn SB I-15 WB I-10 Conn SB I-15 2 300 0.18 RSR 3.01 3
Note: CLR=Continuous Lane Reconstruction; RSR=Random Slab Replacement; ACR=Asphalt Concrete Rehabilitation

slide notes

This is a more detailed analysis of the original alternative. Used to determine:

  • Incentive/Disincentive
  • Number of weekends needed (min/max)
  • Working Days
  • Controlling operations
  • Good check to make sure our staging works, and that the Contractor has sufficient access.

slide 15

Traffic Study (Dynameq)

Closure Study 1 Delay (min) Study 2 Delay (min)
WB10-SB15 5.5 8.4
EB10-SB15 4.1 7.7
SB15-WB10 4.5 72.6
NB15-E/W10 5.8 58.6
EB10-NB15 5.8 8
Reduce SB 15 3
SB15-E/W60 121.4

slide notes

  • Contracted 2 traffic studies for this project. The second evaluated the first.
  • Studies looked at 6 closures
  • Variables hard to predict: Diversion, detour usage (especially in urban area), effects of economy, user learning
  • Reduce SB 15 will not be done.
  • WB10-SB15 – Actual, closed EB-SB as well. Detour? Delay < 10 min
  • EB10-SB15 – See previous.
  • SB15-WB10 – Closed SB-EB at the same time, and 4th St onramp. Delay < 10 min. Detour at Jurupa?
  • NB15-E/W10 – Have not done yet.
  • SB15-E/W60 – Delay < 10 minutes for at least 2 out of 3 of the closures. 1 closure with delays a bit longer, but still under 30 minutes.

slide 16


  • NTP with CA4PRS study: May 2005
  • 30% PS&E: February 2006
  • Draft CA4PRS Report: March 2006
  • CA4PRS study: October 2006
  • 60% PS&E: January 2007
  • Advertise: September 2008
  • Award: January 2009
  • Construction Begins: April 2009
  • Anticipated Completion: December 2010

slide notes


slide 17


slide notes

  • We have finished all of the weekend closures.
  • About 18 closures used, so the Contractor got the full incentive.
  • As is often the case, the Contractor changed many things in the construction staging, and was very aggressive in pursuing the incentive.
  • We also experienced lower traffic delays than anticipated, we believe in part due to the economy, and in part due to our public awareness efforts.

slide 18


Screenshot of the PCCP Deterministic - OntFix project detail screen, which requires user inputs for project description, analyst name, route name, beginning KM, objective/scope, location, and project notes.

slide notes

  • Objective/Scope is changed for each closure.

slide 19


Screenshot of the PCCP Deterministic - OntFix activity constraints screen and a construction window settings popup screen.

slide notes


slide 20


Screenshot of the PCCP Deterministic - OntFix resource profile screen, which requires user inputs for resources including demolition hauling trucks, batch plant, concrete delivery truck, base delivery truck, and paver

slide notes

  • Couple times they had two pavers out there, but I don't believe they ever operated them simultaneously, due to lack of crews.

slide 21


Screenshot of the PCCP Deterministic - OntFix schedule analysis screen, which requires user inputs for construction window, curing time, section profile, working method, change in roadway elevation, and lane widths.

slide notes

  • Note the user-defined section. Need to define the base more, since it was placed in two lifts.
  • Set concurrent double lane/single lane as needed. May need to do portions of a stage separately.
  • Sometimes our outside lane width was 2 feet wider.

slide 22


Screenshot of the PCCP Deterministic - OntFix production details screen, which requires user inputs for a variety of quantities and qualities.

slide notes

  • Note the constraint resource.

slide 23

Inputs: Predicted Vs. Actual

Tab Input Study Actual Unit Notes Default Values
Activity Constraints† Mobilization 3 1 hrs 2-3
Activity Constraints † Demobilization 2 Varies hrs 4-6
Activity Constraints† Concurrent - Demo to Base 15 11 hrs 1-2 (Sequential), 9-11 (Concurrent)
Activity Constraints† Concurrent - Base to PCC 8 5 hrs 1-2 (Sequential), 9-11 (Concurrent)
Resource Profile Demo Hauling Truck 22 22 tonne 22
Resource Profile* Demo Trucks per hour per team 10 10 ea 10 for cut & lift, 12 for impact methods
Resource Profile*† Demo Packing Efficiency 0.5 0.55 % Used impact method instead of cut & lift 0.5 for cut & lift, 0.6 for impact
Resource Profile† Demo Number of Teams 2 3 ea BIG IMPACT! 2
Resource Profile* Demo Team Efficiency 0.7 0.7 %
Resource Profile†† Base Delivery Truck Cap. 10 6 m3 Used end dumps instead of bottom dumps 10 for buttom dump, 6 for end dump
Resource Profile† Base Trucks Per Hour 8 16 ea 10
Resource Profile Base Truck Packing Eff. 100 100 %
Resource Profile* Batch Plant Capacity 90 90 m3/hr 100
Resource Profile Concrete Delivery Truck Cap. 6 6 m3 6-7
Resource Profile Concrete Trucks Per Hour 15 15 ea 15
Resource Profile Concrete Truck Packing Eff. 100 100 %
Resource Profile* Paver Speed 2 2 m/min 2
Resource Profile Number of Pavers 1 1 ea Sometimes double-lane width
Schedule Analysis †† PCC Thickness 290 315 mm 345 mm for mainline lanes (315 for connectors)
Schedule Analysis Base Thickness 152.4 150 mm
* Unable to verify actual value in field
†Difference speeds up actual production
††Difference slows down actual production

slide notes

  • Explain the table headings
  • Got this input from talking to construction inspectors, and reviewing the inspector diaries.
  • Haven't been able to verify every actual input (marked with an asterisk), but we have gotten most of the critical ones.
  • Define orange vs. blue highlighting
  • Contractor has really understood that the demolition operation is often the critical operation.
  • The two slow downs have minimal impact (only minor slow down)
  • Show defaults. Note that at the time of analysis, defaults were not included with the program.

slide 24

Predicted Vs. Actual

  • Random slabs as night work
  • Sometimes paving two lanes wide on connectors
  • Combined stages
  • Concurrent vs. Sequential

slide notes

  • Demo efficiency was validated by how the contractor performed the Super-Slab work: Fort Miller stated that they have seen very few contractors as efficient as this one on the demo operation.

slide 25

Predicted Vs. Actual (PRELIMINARY)

Stage Plan1 Stage Study1 Stage Description No. of Weekends Study2 No. of Weekends Actual2 No. of Weekends Revised Inputs3 Notes
4B,C 5B,C SB I-15 connectors to E/W SR-60 2 2 Contractor was restricted by width of connector, which forced him to pave one lane at a time. Only 2 demo teams used. Thus very similar to study
2B,C 2B,C SB Jurupa offramp, W10-S15 conn, E10-S15 conn 4 2 2 Study had separate closures for 2B, 2C. Contractor chose to combine stages.
2D,E 2D,E SB I-15 connectors to E/W I-10 5 2 2 Contractor may have included more in 2E,F combination, also need to determine how contractor handled 3-lane widths
2E,F 2E,F Fourth St SB ramps 3 1 1 Added 2E work north of S15-E10 connector diverge

Footnotes (Column descriptions):

1"Plan" is the stage designation as it is called out on the project plans. "Study" is the stage designation as it is called out in the design study. Differences exist because of changes that occurred between when the study was completed and the project design was finished.

2"Study" is the number of closures (weekends) estimated to be needed by the design study to complete the work for the stage. "Actual" is the number of closures actually required to complete the work.

3"Revised Inputs" indicates how many closures were estimated to be needed using the revised inputs for CA4PRS shown in the previous slide.

slide notes

  • Read footnotes later.
  • This is PRELIMINARY. Used stages that required more than 1 closure.
  • Explain: This was create merely by running the revised inputs on the same lane-kilometer objective.
  • On the first line, Contractor was restricted by the location to work in a manner very similar to what the study had predicted.
  • Second line, the Contractor combined stages, but the study had them separately.
  • Third line needs more work to determine why the Contractor was able to perform the work in 2 closures. This highlights the challenge of analyzing this type of work. Contractor significantly revised/combined stages to make work more efficient for his crews.

slide 26

Time Spent

  • Pre-Construction Study:
    • ~160 hours
  • Validation of Results:
    • ~40 hours

slide notes

  • Time consuming part is data collection and deciding how the contractor will perform the work.
  • In validation, these things are know already (and hopefully your CA4PRS file is already set up), so validation occurs much more quickly.
  • More to be done on validation.

slide 27

Lessons Learned

  • Construction experience is IMPORTANT
  • Design input important for efficiency
  • Breadth of knowledge required
    • Traffic
    • Pavement
    • Construction
    • Estimating
  • Team approach may be best

slide notes

  • A team approach may be best, maybe even a dedicated team.
  • D8 is considering a team that would also do things like LCCA, estimate verification, etc.
  • Interested to hear suggestions from others with experience with the software.

slide 28

For Further Study

  • More detailed field validation.
  • How much did the economy effect traffic volumes, allowing the combined closures?
  • What kind of traffic diversion did we get?
  • Document experience with precast pavement (Super-Slab) to validate precast module.

slide notes


slide 29

Contact Info

Jonathan den Hartog
(909) 383-5998

More info:

slide notes


Office of Operations