1 Iowa’s Experience with Two-lane to Three-lane Conversions Wen Li, Master’s Student and Dr. Alicia Carriquiry Iowa State University Department of Statistics, Dr. Michael Pawlovich, PE and Tom Welch, PE Iowa DOT Office of Traffic and Safety for the Iowa Department of Transportation Office of Traffic and Safety
2 Preliminary Results Density: 12 percent (reduction) Rate: 34 percent (reduction) NO control sites – so preliminary results – future effort
3 What Converting an existing 2-lane roadway to a 3- lane roadway typically 2-lanes with center lane roadway typically 2-lanes with center continuous left-turn lane continuous left-turn lane Important left-turn lane Also – sometimes bike lanes and pedestrian accommodations
4 What: 3-lane Source: Iowa DOT RoadView Arnold’s Park, IA – US 71 NB
5 What: 3-lane Source: Iowa DOT RoadView Arnold’s Park, IA – US 71 NB
6 What: 3-lane Source: Iowa DOT RoadView Arnold’s Park, IA – US 71 NB
7 What: 3-lane Source: Iowa DOT RoadView Indianola, IA – IA 92 NB
8 What: 3-lane Source: Iowa DOT RoadView Harlan, IA – US 59 NB Speed limit Note: Oncoming van is entering turn lane with vehicle behind it. Speed limit is 45 mph thru the school site. Harlan, IA School
9 History Treatments in Iowa 35 from 1981 – 2004 28 from 1991 – 2004 Studies 2005 – Florida DOT
10 Bayesian Before/After Objective: Assess whether 2-lane to 3-lane conversions appear to result in crash conversions appear to result in crash reductions on Iowa roads reductions on Iowa roads
11 Bayesian Before/After Data 34 treatment – 1 dropped as treatment in 1981 25 years ( ) Volumes: 2,500 to 15,700 ( ) Currently NO control sites – future effort
12 Site Data Site IDCityPopulation 2000Volume Ames50,7316,800 2Ames50,7317,900 3Ames50,7317,500 4Ames50,7318,500 5Ames50,7315,000 6Ames50,73112,200 7Ames50,73110,000 8Arnolds Park1,16215,700 9Atlantic7,2578,000 10Council Bluffs58,2688,200 11Creston7,5975,300 12Davenport98,3599,500 13Davenport98,3598,500 14Dyersville4,0359,000 15Eldridge4,1597,100 16Harlan5,2825,300 17Hull1,9604,400 18Indianola12,99813,500 Site IDCityPopulation 2000Volume Iowa City62,22010,000 20Iowa City62,2207,500 21Manchester5,2577,900 22Manchester5,2577,500 23Maquoketa6,11210,000 24Mason City29,1723,200 25Mason City29,1727,400 26Nevada6,6583,400 27Nevada6,6585,600 28Nevada6,6584,000 29New London1,9372,500 30Newton15,57911,500 31North Liberty5,36711,500 32Oelwein6,6925,000 33Orange City5,5824,500 34Pella9,8323,200 35Waverly8,9686,000 Note: Maquoketa dropped from analysis (1981)
13 Site Data Site IDCityLength (mi) Completion Year 1Ames Ames Ames Ames Ames Ames Ames Arnolds Park Atlantic Council Bluffs Creston Davenport Davenport Dyersville Eldridge Harlan Hull Indianola Site IDCityLength (mi) Completion Year 19Iowa City Iowa City Manchester Manchester Maquoketa Mason City Mason City Nevada Nevada Nevada New London Newton North Liberty Oelwein Orange City Pella Waverly Note: Maquoketa dropped from analysis (1981)
14 Monthly Crash Densities and Rates Site 4 Intervention Density Rate
15 Modeling/Statistics Crashes: random, rare count events Model: Poisson with/ log mean expressed as a piece-wise linear function of: - time period (months) - time period (months) - seasonal effects - seasonal effects - random effect for each site - random effect for each site Adopted a Bayesian statistical framework for estimation of model parameters estimation of model parameters Model essentially the same as for prior 4-lane to 3- lane analysis lane analysis
16 Preliminary Results/Discussion 12% (reduction) in crashes/mile 34% (reduction) in crash rate Recall NO control sites – so preliminary results Recall NO control sites – so preliminary results
17 Results/Discussion Prior Florida DOT study: 32 percent decline in crashes not “significant” at 95 percent confidence level Need to include control sites - future
18 Questions? & Discussion