Wednesday, July 11, 2012

US and European rail safety

Awhile back, Alon Levy posted about comparative rail safety, which showed the US to be the least safe nation in the world for passenger rail. Unfortunately, it is sourced from Wikipedia in large part and is focused strictly on passenger safety involving fatalities. What about the rail systems as a whole however? Is the US simply an outlier due to what is promoted as the best freight rail system in the world resulting in a lack of passenger rail focus and service? One might argue that the relative safety is the same, but that rest of the world has sufficiently more people riding trains that there's only an apparent increase in safety (such as one might see if two trains existed, both of which ran the same number of miles and suffered the exact same accidents, but one had four times as many passengers as the others; all other things being equal, the second train would appear four times safer than the first on a per passenger-mile basis).

European data is sourced from this document, starting with the tables on page 45. US data comes from this extremely handy data generator courtesy of the FRA. The only data used for the US is train accidents and highway/rail accidents. Potentially this may result in an undercounting compared to Europe.

This isn't entirely comparable to Alon's post, of course. I'm only examining the year 2010 rather than a spectrum of years and I'm looking at overall safety, not just passenger fatalities.

Charts and data lie beneath the page break






Country
Total Accidents
Millions of Train-KM
Accidents/Million-Train-KM
AT
79
156
0.51
BE
40
98
0.41
BG
39
31
1.26
CT
1
6
0.17
CZ
125
160
0.78
DE
297
1032
0.29
DK
22
83
0.27
EE
31
9
3.44
EL
39
17
2.29
ES
45
187
0.24
FI
23
51
0.45
FR
155
485
0.32
HU
142
97
1.46
IE
3
18
0.17
IT
103
324
0.32
LT
44
14
3.14
LU
3
8
0.38
LV
41
17
2.41
NL
24
146
0.16
NO
20
46
0.43
PL
449
219
2.05
PT
42
40
1.05
RO
271
94
2.88
SE
69
141
0.49
SI
21
19
1.11
SK
231
48
4.81
UK
62
520
0.12
Total
2421
4066
0.60
USA
3,917
1134
3.45
AMTRAK
224
65
3.45
BNSF
742
272
2.73
CN
188
28
6.71
CP
69
22
3.14
CSX
591
151
3.91
KCS
97
17
5.71
NS
494
147
3.36
UP
825
258
3.20




What about passenger fatalities?



Passenger Fatalities
Millions of Passenger Kilometers
Fatalities per Billion Passenger Kilometer
AT
0
10700
0.0000
BE
18
10493
1.7154
BG
0
2100
0.0000
CT
0
497
0.0000
CZ
2
6553
0.3052
DE
0
83702
0.0000
DK
0
6587
0.0000
EE
0
456
0.0000
EL
1
1144
0.8741
ES
15
20978
0.7150
FI
0
3959
0.0000
FR
2
81750
0.0245
HU
3
7666
0.3913
IE
0
1678
0.0000
IT
7
43474
0.1610
LT
0
373
0.0000
LU
0
350
0.0000
LV
0
741
0.0000
NL
0
16621
0.0000
NO
0
3153
0.0000
PL
7
17800
0.3933
PT
1
4111
0.2432
RO
4
5500
0.7273
SE
2
11036
0.1812
SI
0
813
0.0000
SK
0
2291
0.0000
UK
0
55831
0.0000
Total
62
400357
0.1549
USA
3
31872
0.0941
AMTRAK
2
10331
0.1936

On the face of it, the US actually appears to be safer than the Europeans for fatalities, but I believe this is a statistical anomaly due to the extreme safety of passenger rail travel. The number of fatalities are so low that any death, especially in the event of a mass casualty event such as the Halle train collision in Belgium (responsible for all 18 of their fatalities in 2010), that it causes a major swing in the relative data. The infrequent fatality rate and the tendency of passenger fatalities to be restricted to one or two major events suggests that the European standard of focusing on crash prevention rather than increasingly higher crash safety standards is the better approach.

What is surprising is that the US has a much higher rate of overall accidents. Given that even a simple derailment can be expensive, especially when it blocks a major route for a significant amount of time, one has to wonder why the US has such a high rate of accidents. One answer may be that the US has significantly more grade crossings, both in absolute and relative numbers, than do European nations and that this is responsible for the higher rate of accidents. Slightly more than half of all US accidents were highway/rail incidents while only a quarter of European accidents were level crossing accidents. Another issue is that of differences in reporting: Different threshold reporting requirements, or European figures not including yard accidents while the US figures do (I don't know if the European figures do exclude yard accidents), could result in mismatched data sets. Collisions and derailments should have similar reporting rates across nations however.


Country
Collisions
Millions of Train-KM
Collisions per Million Train Kilometers

AT
3
156
0.0192

BE
5
98
0.0510

BG
2
31
0.0645

CT
0
6
0.0000

CZ
3
160
0.0188

DE
13
1032
0.0126

DK
1
83
0.0120

EE
1
9
0.1111

EL
4
17
0.2353

ES
2
187
0.0107

FI
0
51
0.0000

FR
15
485
0.0309

HU
1
97
0.0103

IE
0
18
0.0000

IT
2
324
0.0062

LT
0
14
0.0000

LU
1
8
0.1250

LV
1
17
0.0588

NL
5
146
0.0342

NO
9
46
0.1957

PL
4
219
0.0183

PT
2
40
0.0500

RO
10
94
0.1064

SE
3
141
0.0213

SI
0
19
0.0000

SK
13
48
0.2708

UK
8
520
0.0154

Total
108
4066
0.0266

USA
131
1134
0.1155

AMTRAK
6
65
0.0923

BNSF
27
272
0.0993

CN
7
28
0.2500

CP
3
22
0.1364

CSX
23
151
0.1523

KCS
3
17
0.1765

NS
16
147
0.1088

UP
23
258
0.0891



Country
Derailments
Millions of Train-KM
Derailments per Million Train Kilometers

AT
2
156
0.0128

BE
2
98
0.0204

BG
1
31
0.0323

CT
0
6
0.0000

CZ
3
160
0.0188

DE
19
1032
0.0184

DK
1
83
0.0120

EE
0
9
0.0000

EL
2
17
0.1176

ES
7
187
0.0374

FI
1
51
0.0196

FR
14
485
0.0289

HU
1
97
0.0103

IE
0
18
0.0000

IT
3
324
0.0093

LT
1
14
0.0714

LU
0
8
0.0000

LV
0
17
0.0000

NL
3
146
0.0205

NO
4
46
0.0870

PL
17
219
0.0776

PT
3
40
0.0750

RO
0
94
0.0000

SE
7
141
0.0496

SI
0
19
0.0000

SK
2
48
0.0417

UK
6
520
0.0115

Total
99
4066
0.0243

USA
1339
1134
1.1808

AMTRAK
21
65
0.3231

BNSF
334
272
1.2279

CN
47
28
1.6786

CP
24
22
1.0909

CSX
175
151
1.1589

KCS
31
17
1.8235

NS
156
147
1.0612

UP
341
258
1.3217


Potentially, there may still be some oddness with reporting when it comes to derailments: Even though it is the lowest of American railroads, Amtrak reports a derailment higher an order of magnitude higher than European railways, and American freight railroads are all significantly higher.

I'll admit that I can't really think of a reason why collisions should have significantly different reporting rates between Europe and America, though it is possible with sideswipes and the like. Be that as it may, the data does show that the US has a significantly higher accident rate than Europe, assuming a fairly comparable set of reporting requirements, though with a passenger fatality rate significantly lower than what the accident rate would suggest (on par with Europe in 2010). With a suggested total cost of half a million euros per accident, among all accidents, improving our safety, both for freight and passenger rail, would greatly benefit America's economy, to the sum of approximately $2 billion annually.



2 comments:

  1. Do you have the same figures stretching back a few years? This could smooth trends for passenger fatalities. For example we know the US rate is about 8 per year if you take a 20-year average and only look at the accidents on the Wikipedia list, and this is enough to put the US above Europe by some margin.

    ReplyDelete
    Replies
    1. Crud, didn't realize that the graphs didn't show. Worked in preview. Makes the next post look rather dumb.

      Anyhow, I've pulled passenger fatality, injuries, and passenger-miles back to 1976 (bear in mind that it is occasionally screwy). I can pull the other info as well if need be, but I believe that's what you were wanting?

      Delete