1In a previous paper [Dobruszkes et al., 2024], we investigated the inclusiveness of public transport (PT) in Brussels at the stop level; that is, whether these stops are accessible to disabled persons with physical, visual or cognitive impairment. Based on nine travel constraints, the results showed that a significant proportion of PT stops – i.e. surface stops and underground stations – are not designed appropriately. Policies and PT design have improved (including lifts in most underground stations, more low-floor trams and about 70 bus/tram stops being upgraded every year), but the legacy of decades of negligence remains significantly visible.
- 1 Not talking about the design of public space and the resulting easiness of access to/egress from PT (...)
2However, PT stops should only be considered as a first level of investigation. Specifically, any person making a journey through the city is travelling from an origin to a destination and possibly via intermediate stops; hence the rationale for also investigating the inclusiveness of PT at the itinerary level. Shifting the focus to itineraries would provide further insights because for disabled people a PT journey is possible only if the departure stop, intermediate stops (if any), arrival stop and the vehicles are designed accordingly1. This is even more challenging than when looking at the design of each PT stop in isolation.
3In this context, the aim of this paper is to assess the gap in PT access to the city faced by travellers with specific needs compared to travellers who do not encounter any regular obstacles when using public transport within the Brussel-Capital Region (BCR). By PT accessibility, we mean the ease with which journeys can be made. Our objective is twofold. First, we intend to provide new scientific evidence on the extent of the gap in PT access to the city disabled persons face, considering Brussels as a case study. Second, we hope to contribute to the public debate and influence policies and practices related to inclusive PT in cities in general and in Brussels in particular.
4As its title suggests, this paper is the direct continuation of our research at the stop level. Key discussions on the Brussels context, gaps in the academic literature, our specific approach based on travel constraints rather than specific disability categories, and the appropriate vocabulary in disability studies can be found in Dobruszkes et al. [2024]. The remainder of the paper is as follows. The next section proposes a brief literature review to highlight the lack of quantitative research on accessibility for disabled persons. Section 2 details the methodology and data used. Section 3 presents the results, and Section 4 concludes.
5First, it seems research on disability is more interested in employment and education than in transport and accessibility issues, notably in the US [Harris et al., 2014]. However, even though scholars have focused on transport/mobility issues, gaps in PT access to the city for disabled persons have been investigated mostly through qualitative studies, which, by definition, cannot present a global picture. A major reason for this is that the field of disability studies is dominated by qualitative approaches, if not by the rejection of quantitative perspectives, as evidenced by the content of the flagship journal, Disability Studies Quarterly [Blanchard et al., 2025], and several books [e.g., Watson et al., 2012; Egard et al., 2022].
6This qualitative dominance would be due to several factors discussed by Blanchard et al. [2025] and includes a rejection of quantitative classifications of human ability and performance, which in several cases has led to scientific racism and eugenic policies. In addition, the early political struggles of the disability rights movement against discrimination did not require quantitative assessments. The main issue at the time was (and is still, to some extent), “establishing the humanity of disabled people – insofar as shared humanity is considered sufficient to obtain civil rights” [Blanchard et al., 2025].
7In contrast, we argue that in a modern society where some progress and rights have been secured, quantitative assessments of the various difficulties disabled persons face in their daily lives can help make policymakers and the public aware of the issues at stake. The results of quantitative assessments can also help to legitimise the claims of disabled people.
8Furthermore, academic studies on disability have neglected the geographical (aka spatial) dimension of disability. As Imrie [2000] notes, there is a “spatial amnesia” and
This is curious because geography is fundamental to an understanding of the social, economic, and political opportunities and/or constraints underpinning the lives of disabled people.
9This “spatial amnesia” is also strange when one considers that:
many disabled people have little option but to stay at home because public transport is too poor to facilitate their mobility. This limits their geographical boundaries, so preventing access to a range of places and associated goods and services. In all of these senses, geography matters [Imrie, 2000].
10One can turn the reflection around and ask to what extent quantitative research works on urban accessibility have taken account of disability. Indeed, the calculation of PT accessibility across a city has become somewhat commonplace in the fields of transport geography, transport planning and transport studies in general. This has been driven particularly by advances in open and standardised data, as well as in available software, including free and open-source options [Higgins et al., 2022].
11However, scholars have widely and implicitly assumed any individual can board or alight from any vehicle [see, e.g., Dobruszkes, 1995; Lebrun, 2018b; Leclercq et al., 2015]. Such research thus neglects travellers with specific needs and the fact that inappropriate PT design (both stops and vehicles) does prevent disabled persons from using public transport. However, the few pieces of research that have quantitatively analysed PT accessibility mediated by disability have found a significant gap [see, e.g., Qatra, 2016; Liu et al., 2023]. Our paper complements these works by considering different impairments and the case of Brussels. As the capital of Belgium, Brussels has a dense and diverse PT network consisting of underground lines, trams operating underground (“premetro”) and at ground level, and bus lines. The network is a mix of inclusive and non-inclusive facilities and vehicles [Dobruszkes et al., 2024], which makes Brussels a good case study.
12Our research strategy is based essentially on a comparison between:
-
Travel conditions for a wide range of intra-urban journeys for travellers who do not face any particular obstacles in using public transport (hereafter “unconstrained accessibility”);
-
And the conditions for the same journeys for travellers facing different travel constraints due to some specific needs (hereafter “constrained accessibility”).
13Travel conditions will be estimated by trip modelling fed by a PT network, PT timetables and public spaces. The difference between the two measures (unconstrained or constrained accessibility) will provide information on the magnitude of the gap in PT access to the city faced by disabled people. Tables will present the mean results, and results disaggregated at the PT stop level will be mapped.
14It is worth noting that, we, as authors, do not face major obstacles in using PT, so we may misunderstand the relevant indicators to be studied as well as the main travel constraints disabled persons face when using (or intending to use) PT. To avoid such misunderstanding, our quantitative approach is supported by rich qualitative input, including extensive talks with NGOs and public services (PT operators, in particular) and 33 PT journeys we undertook with 24 disabled people. These inputs helped us confirm our methodology of working on the basis of travel constraints but also led us to modify some constraints to understand the limits of the modelling and finally to analyse the results in a better-informed way (see Appendix 1).
15People with impairments have a diversity of profiles and needs, so it is futile to define typical profiles. Instead, it would be more appropriate to consider a range of nine constraints that focus on the concrete needs of people with impairments, including physical, visual and cognitive impairments. In this paper, we reuse a subset of six travel constraints (Table 1, please refer to Dobruszkes et al. [2024] for more detailed information).
Table 1. The travel constraints considered
Travel constraint
|
Description
|
Hard step-free
|
Need to embark/disembark from a stop with an absence of a step and a gap between vehicles and platform edges.
|
Soft step-free
|
Similar to “hard step-free”, but here, a “slight” gap between vehicles and platform edges can be overcome.
|
No stair/no escalator
|
Need to avoid stairs and escalator.
|
Tactile paving
|
Need to embark/disembark from a stop with tactile paving.
|
No complex station
|
Need to restrict oneself to the surface network.
|
Surface
|
Only surface stops are selected for the route calculation.
|
16Journey modelling is done by estimating the most plausible itinerary between given origins and destinations. The investigation is explicitly limited to urban PT travel and to the STIB-MIVB network. We acknowledge that the interregional Flemish (De Lijn) and Walloon (TEC) bus services as well as the national railway services contribute to some extent to improving access to the city. However, the number of districts with improved PT access (understood as travel time) to other parts of the city thanks to these three operators is rather limited [Lebrun, 2018b]. Moreover, the BCR has very little influence on these services.
17Trips originate from all STIB-MIVB stops that were active during daytime in October 2022, excluding those stops temporarily not served due to public works. Stops serviced only by overnight services or in the case of special events (e.g., when parks are closed and EU summits) are also excluded. There are, therefore, 2 174 departure points throughout the city.
18The destinations comprise a set of 20 points of interest (POIs). We tried to ensure diversity in terms of both geographical location and travel purposes (from business districts to shopping to academic hospitals to leisure spots to sport facilities to the prison) (Figure 1). These choices have been discussed with NGOs and public services.
Figure 1. The 20 destinations
19Between the 2 174 origins and 20 destinations (i.e. 43 480 itineraries), travellers will have to walk in public spaces and use public transport. Public spaces are known thanks to OpenStreetMap (OSM). As for the STIB-MIVB network and services, they are freely available through files that follow the GTFS format. We used the network and timetable as it was at the beginning of October 2022.
20OSM and GTFS data were used to feed OpenTripPlanner (OTP) 1.5. OTP is a free and open-source software designed to estimate itineraries within a transport network. A number of parameters had to be set to calibrate the model, including walking speed, reluctance to walk, maximum walking distance and minimal transfer time2. The route eventually selected by OTP to represent the journey is the one that minimises the journey time, given the parameters’ value. Note that journeys starting the day before were not accepted.
21In a first step, we asked OTP to calculate the 43 480 itineraries, taking into account all stops and routes (unconstrained accessibility). In a second step, we asked OTP to recalculate the 43 480 itineraries, but taking into account each of the six travel constraints introduced above (constrained accessibility). Any stop and any route (based on the vehicle type assigned to the service) that did not meet a given travel constraint under consideration was excluded. In other words, only the combination of suitable stops and suitable vehicles was still considered for inclusion in the calculations. If the rolling stock was heterogeneous and mixed suitable and unsuitable vehicles for the constraint considered (e.g., on tram route 81), we excluded the whole route from the calculations, as disabled persons could not wait at a stop for the suitable vehicle to arrive especially as this information is not currently available to travellers.
22For some constrained accessibilities (Hard step-free, Soft step-free and Tactile paving), we also imposed an upper threshold of 500 metres for the walking distance. Without such threshold and in the case of many excluded stops, OTP estimates itineraries with very long walking distances (up to several kilometres). This is absurd in the absolute and even more so when it comes to journeys made by disabled people. Setting a maximum walking distance avoids this problem. In addition, the qualitative analysis highlighted the concern for an inclusive public space, confirming the usefulness of limiting this distance. By consequence, if the route involves a walking distance of more than 500 metres between the point of departure and an alternative point of departure, between two stops (in case of transfer) or between the closest stop to the POI and the POI itself, the departure point then has no valid solution and is excluded (it will appear as such in maps).
- 3 Note that we tested whether imposing alternative but close arrival times (9:03, 9:07, 9:11) affecte (...)
- 4 For instance, tram 51 was then diverted via Brugmann Avenue instead of Alsemberg Rd.
23All calculations were made for an arrival at the POIs on Tuesday 4 October 2022 by 9:003. Any later changes to the network, timetable, vehicles used and stop’s design are therefore not included in our results4.
24Finally, the 33 PT journeys with disabled people highlighted that strict compliance with the identified travel constraints are not always observed during the rides. On the one hand, our calculations could underestimate the gap in PT access to the city faced by disabled persons. Indeed, the modelling assumes that accessible stops and vehicles are actually accessible. In the real world, local/temporary circumstances may prevent a disabled person from boarding or alighting. As our qualitative analyses have highlighted, causes may include, for instance: walkability issues from/to the PT stop bus that stopped away from the stop (before/after the stop line and/or away from the platform’s edge) due to obstacles or driver’s behaviour; bus ramp or lift being out of service; etc. In addition, our modelling supposes that public spaces are truly accessible, which is not the case in the real world [Levine, 2024]. On the other hand, the degree of autonomy varies across disabled persons even if they have the same impairment or use a similar mobility device. Therefore, our calculations may also sometimes overestimate the gap in PT access to the city for disabled people who manage to overcome the travel constraints, thanks to their skills, their strength or someone’s help.
25Among the various outputs offered by OTP, we have considered the following four metrics based on the qualitative part of this research:
-
The number of departure stops. It is important to note that this does not mean the proportion of appropriate stops. Indeed, if a stop is excluded because it does not meet a given constraint, OTP will seek an alternative stop within 500 metres. To obtain the share of inclusive stops for each constraint, see Dobruszkes et al. [2024].
-
Total travel time between a departing stop and a POI (in minutes): A basic metric. Travellers are expected to value shorter journeys, all other things being equal.
-
Number of transfers: Transfers are generally known to be painful, so PT users may opt for longer but direct journeys [Iseki and Taylor, 2009; Dobruszkes et al., 2011; Garcia-Martinez et al., 2018]. Our qualitative analyses have shown that transfers are to be even more painful for disabled people because each boarding and alighting is a potential source of inconvenience, stress and discomfort.
-
- 5 Since our departing points are all the PT stops, there is a walk at the origin only when a stop is (...)
Walking distance: This is the sum of “walking” (including in wheelchairs) before the departure stop if any5, during transfers and between the final stop and the POI. For disabled persons, the shorter the distance, the better it is. Indeed, having to “walk” increases the risk of encountering barriers to movement.
26We are aware that these four metrics may mask diverging situations among travellers and that there may be a gap between estimated travel constraints and their actual perception by individuals, as demonstrated by Ryan and Pereira [2021]. We understand that this is the price of providing a spatially comprehensive analysis.
- 6 No-constraint accessibility, then the six constraints.
- 7 Even not counting the maps with all the possible combinations of travel constraints.
- 8 https://zenodo.org/records/14623247
27Of course, we cannot show all our results. In particular, 420 maps (20 POIs x 7 situations6 x 3 indicators) were generated automatically7. We therefore start with global results, then focus on results per destination and per origin-destination pair for the step-free and tactile paving constraints. The decision to focus on these travel constraints is twofold. On the one hand, they have the greatest impact on the network, as many stops do not meet the criteria for being considered accessible. On the other hand, they are among the most common travel constraints encountered during the 33 PT journeys. For the rest, the 420 maps will be made available as online supplementary material through a free repository8.
28Table 1 unveils the overall results, with the 20 destinations considered together. It shows that imposing a travel constraint downgrades accessibility with a wide range of magnitudes. The lowest impacts result from “No stairs, no escalators” and “No complex station”, which makes sense since all surface stops meet these constraints but also most underground stations. The “Surface” constraint results in longer journey times (+21 % on average) and more transfers (+20 %), but at least all 2 174 departure points remain included.
- 9 New M7 train sets delivered or being delivered fix this issue (provided the platform is 100 cm abov (...)
29In contrast, the worst results are found for “Tactile paving” and “Hard step-free”. “Tactile paving” implies a 30 % drop in departure stops, longer journeys (+49 %) with more transfers (+61 %) and more walking (+54 %). It should be remembered that the 30 % drop in departure points does not mean that 70 % of the stops meet the constraint. The proportion is actually 34 % [Dobruszkes et al., 2024]. The difference between 34 % and 70 % is because the model searches for alternative stops within the 500m walking distance threshold, when necessary. As for the “hard step-free” constraint – which assumes autonomous step-free accessibility for any wheelchair user without assistance, regardless of their force and ability, one finds a 26 % reduction in departure stops, much longer journeys (+68 %), much more transfers (+70 %) and longer “walks” (+38 %). Let us recall that the entire underground network does not meet the criteria because of the systematic gap between platforms and trains at the time of our data (not to mention lifts still missing in several stations)9.
30Logically, the soft step-free constraint gives better results than the “hard step-free” constraint. But interestingly, results for “soft step-free” are much closer to unconstrained accessibility than to “hard step-free”. The gap between hard and soft step-free highlights the different levels of difficulty faced by disabled people.
Table 2. The overall impact of travel constraints on accessibility
|
Unconstrained accessibility
|
Constrained accessibility
|
|
No constraint
|
Soft step-free (500 m)
|
Hard step-free (500 m)
|
No stairs, no escalators
|
Tactile paving (500 m)
|
No complex station
|
Surface
|
Departure stops (n)
|
2 174
|
1 948
|
1 612
|
2 174
|
1 519
|
2 174
|
2 174
|
|
|
–10 %
|
–26 %
|
|
–30 %
|
|
|
Travel time (minutes)
|
45
|
50
|
76
|
46
|
68
|
48
|
55
|
|
|
+10 %
|
+68 %
|
+2 %
|
+49 %
|
+6 %
|
+21 %
|
Transfers (n)
|
1,2
|
1,3
|
2,0
|
1,2
|
1,9
|
1,2
|
1,4
|
|
|
+13 %
|
+70 %
|
+1 %
|
+61 %
|
+5 %
|
+20 %
|
Walking distance (m)
|
427
|
530
|
590
|
430
|
657
|
434
|
410
|
|
|
+24 %
|
+38 %
|
+1 %
|
+54 %
|
+2 %
|
–4 %
|
Average results of mean values calculated for each of the 20 points of interest.
“500 m” means a maximum walking distance of 500 m between PT stops and between the final stop and the destination.
31This section presents the results disaggregated by destination. For the sake of clarity, three key travel constraints (soft step-free, hard step-free and tactile paving) are considered and compared to unconstrained accessibility.
32Overall, the reduction in the number of departure points (Table 3) is rather homogeneous, except for four destinations (ADEPS, Bourget, Prison Haren and Stalle/Neerstalle) where it drops to very low values for at least one of the three travel constraints considered here. This happens when neither the arrival stop nor the surrounding stops meet a specific travel constraint. It is then impossible to reach the destination by PT. Trips are then possible only when walking from other nearby stops within 500 metres, hence the very low number of departure points. It is worth noting that such a massive exclusion of departure points occurs only for some peripheral destinations (although not for all peripheral destinations), where PT network density is lower and so is the probability of finding an alternative stop. In the specific case of Stalle/Neerstalle destination, the problem seems to have been aggravated by major infrastructure works in the adjacent district, so that travellers had to board or alight at non-inclusive temporary stops at the time. The non-inclusive nature of temporary stops is a pending issue in Brussels, despite the emerging use of more inclusive solutions for long-term disturbances (removable temporary structures).
Table 3. Results by destination: Number of departure points
Destination
|
No constraint
|
Soft step-free (500 m)
|
Hard step-free (500 m)
|
Tactile paving (500 m)
|
ADEPS
|
2 174
|
2 051
|
1 786
|
2
|
Blue Tower
|
2 174
|
2 051
|
1 801
|
1 904
|
Bourget
|
2 174
|
2 049
|
1 787
|
7
|
Bourse
|
2 174
|
2 051
|
1 801
|
1 907
|
Central station
|
2 174
|
2 051
|
1 789
|
1 907
|
Erasme
|
2 174
|
2 050
|
1 793
|
1 907
|
Flagey
|
2 174
|
2 051
|
1 801
|
1 904
|
Luxembourg
|
2 174
|
2 051
|
1 790
|
1 907
|
Midi station
|
2 174
|
2 051
|
1 779
|
1 898
|
Nord station
|
2 174
|
2 051
|
1 769
|
1 901
|
Prison Haren
|
2 174
|
1
|
1
|
1
|
Rogier
|
2 174
|
2 051
|
1 796
|
1 905
|
Schuman
|
2 174
|
2 051
|
1 810
|
1 907
|
Stalle/Neerstalle
|
2 174
|
2 049
|
18
|
1 899
|
Tour & Taxis
|
2 174
|
2 051
|
1 783
|
1 879
|
UCL Saint-Luc
|
2 174
|
2 051
|
1 765
|
1 861
|
ULB Solbosch
|
2 174
|
2 051
|
1 810
|
1 899
|
UZ Brussel
|
2 174
|
2 049
|
1 784
|
1 895
|
Westland
|
2 174
|
2 049
|
1 791
|
6
|
Woluwe Shopping
|
2 174
|
2 051
|
1 794
|
1 889
|
Weighted average
|
2 174
|
1 948
|
1 612
|
1 519
|
Gap to no constraint
|
|
- 10 %
|
- 26 %
|
- 30 %
|
Table 4. Results by destination: Mean travel time from all departure points
Destination
|
No constraint
|
Soft step-free (500 m)
|
Hard step-free (500 m)
|
Tactile paving (500 m)
|
ADEPS
|
54
|
59
|
90
|
|
Blue Tower
|
40
|
48
|
57
|
60
|
Bourget
|
52
|
63
|
105
|
|
Bourse
|
34
|
40
|
61
|
59
|
Central station
|
35
|
39
|
64
|
63
|
Erasme
|
52
|
57
|
90
|
68
|
Flagey
|
39
|
43
|
55
|
52
|
Luxembourg
|
37
|
41
|
52
|
49
|
Midi station
|
34
|
40
|
94
|
73
|
Nord station
|
39
|
44
|
70
|
65
|
Prison Haren
|
70
|
|
|
|
Rogier
|
36
|
41
|
62
|
59
|
Schuman
|
37
|
40
|
57
|
51
|
Stalle/Neerstalle
|
50
|
60
|
|
72
|
Tour & Taxis
|
42
|
48
|
65
|
61
|
UCL Saint-Luc
|
51
|
58
|
111
|
105
|
ULB Solbosch
|
47
|
55
|
106
|
74
|
UZ Brussel
|
57
|
61
|
82
|
84
|
Westland
|
58
|
62
|
77
|
|
Woluwe Shopping
|
42
|
48
|
78
|
86
|
Weighted average
|
45
|
50
|
76
|
68
|
Gap to no constraint
|
|
+10 %
|
+68 %
|
+49 %
|
Empty cells when less than 20 departure points (not significant).
“500 m” means a maximum walking distance of 500 m between PT and between the final stop and the destination.
Table 5. Results by destination: Mean number of transfers from all departure points
Destination
|
No constraint
|
Soft step-free (500 m)
|
Hard step-free (500 m)
|
Tactile paving (500 m)
|
ADEPS
|
1,9
|
2,0
|
2,8
|
|
Blue Tower
|
1,1
|
1,4
|
1,8
|
1,5
|
Bourget
|
1,5
|
1,7
|
2,1
|
|
Bourse
|
0,8
|
0,9
|
1,3
|
1,4
|
Central station
|
0,8
|
0,9
|
1,3
|
1,4
|
Erasme
|
1,2
|
1,3
|
2,4
|
1,8
|
Flagey
|
1,0
|
1,2
|
1,7
|
1,6
|
Luxembourg
|
1,1
|
1,2
|
1,5
|
1,4
|
Midi station
|
0,8
|
1,1
|
2,4
|
2,2
|
Nord station
|
0,8
|
1,0
|
2,1
|
1,9
|
Prison Haren
|
1,5
|
|
|
|
Rogier
|
0,8
|
1,0
|
1,8
|
1,8
|
Schuman
|
0,8
|
0,9
|
1,4
|
1,4
|
Stalle/Neerstalle
|
1,3
|
1,5
|
|
2,0
|
Tour & Taxis
|
1,3
|
1,5
|
2,0
|
1,8
|
UCL Saint-Luc
|
1,2
|
1,4
|
2,9
|
2,9
|
ULB Solbosch
|
1,2
|
1,6
|
2,0
|
2,3
|
UZ Brussel
|
1,5
|
1,7
|
2,1
|
2,1
|
Westland
|
1,3
|
1,3
|
2,1
|
|
Woluwe Shopping
|
1,1
|
1,3
|
2,2
|
2,5
|
Weighted average
|
1,2
|
1,3
|
2,0
|
1,9
|
Gap to no constraint
|
|
+13 %
|
+70 %
|
+61 %
|
Empty cells when less than 20 departure points (not significant).
“500 m” means a maximum walking distance of 500 m between PT stops and between the final stop and the destination.
Table 6. Results by destination: Mean walking distance from all departure points
Destination
|
No constraint
|
Soft step-free (500 m)
|
Hard step-free (500 m)
|
Tactile paving (500 m)
|
ADEPS
|
367
|
468
|
470
|
|
Blue Tower
|
287
|
352
|
391
|
845
|
Bourget
|
274
|
489
|
472
|
|
Bourse
|
455
|
606
|
636
|
643
|
Central station
|
373
|
449
|
587
|
710
|
Erasme
|
509
|
596
|
751
|
688
|
Flagey
|
289
|
375
|
402
|
483
|
Luxembourg
|
357
|
433
|
443
|
497
|
Midi station
|
401
|
526
|
816
|
640
|
Nord station
|
438
|
562
|
848
|
781
|
Prison Haren
|
774
|
|
|
|
Rogier
|
398
|
509
|
569
|
579
|
Schuman
|
427
|
510
|
503
|
606
|
Stalle/Neerstalle
|
313
|
467
|
|
695
|
Tour & Taxis
|
352
|
454
|
506
|
518
|
UCL Saint-Luc
|
752
|
1 004
|
666
|
962
|
ULB Solbosch
|
381
|
530
|
673
|
649
|
UZ Brussel
|
335
|
450
|
510
|
497
|
Westland
|
660
|
772
|
704
|
|
Woluwe Shopping
|
405
|
509
|
679
|
730
|
Weighted average
|
427
|
530
|
590
|
657
|
Gap to no constraint
|
|
+24 %
|
+38 %
|
+54 %
|
Empty cells when less than 20 departure points (not significant).
“500 m” means a maximum walking distance of 500 m between PT stops and between the final stop and the destination.
33In contrast, the results for travel time (Table 4), number of transfers (Table 5) and walking distance (Table 6) show a much higher diversity depending on the destination. For instance, the mean travel time under the hard step-free constraint ranges from 52 minutes to the Luxembourg train station to 111 minutes to the UCL Saint-Luc campus and academic hospital. In most cases the results for ‘hard step-free’ are even worse than for “tactile paving”, while the results for the soft step-free constraint are rather close to the no-constraint accessibility.
34It is also noticeable that the ranking of destinations’ accessibility changes when a travel constraint is taken into account. For instance, the lowest mean travel time without constraint is to Bourse (34 minutes), and the highest is to Prison Haren (70 minutes). When “tactile paving” is considered, the lowest mean travel time is found for Luxembourg (49 minutes), Bourse is ranked in fourth place (59 minutes) and UCL Saint-Luc is again the worst (105 minutes), not talking the four destinations that are not significant, given the absence of any PT option.
35It also appears that for “hard step-free” and “tactile paving”, travel time, transfers and walking distance tend to be more aggravated for peripheral destinations (ADEPS, Bourget, Erasme, UCL Saint-Luc, ULB Solbosch, UZ Brussels and Woluwe Shopping), although some central locations also show a sharp deterioration in accessibility. This is particularly the case for journeys to Midi and Nord train stations, where both underground and surface stops are poorly inclusive.
36This section presents selected results in maps disaggregated at the departure stop level (n= 2 174). To facilitate comparisons, all maps have been designed with exactly the same classes and colours, from dark green for the best cases (shortest travel times or lowest number of transfers) to dark red for the worst ones (longest travel times or highest number of transfers).
37Let us first consider “Bourse” as a destination (Figure 2). Bourse (named after the former stock exchange) is a very central district that combines various urban functions, such as shopping, leisure, housing and tourist amenities. Due to its centrality, the area is served by a wide range of PT routes going in all geographical directions. As a result, unconstrained travel times are rather short (often less than 45 minutes), hence the very green map.
38When considering the soft step-free constraint, travel times increase moderately in most cases and several areas are excluded (recall that this means stops and/or rolling stock are not inclusive, and that the model, in this case, does not find an alternative departure stop within 500 metres).
39In contrast, the hard step-free constraint means much longer journeys. For most of the outer districts, travel times exceed 60 minutes, if not 75 or even 90 minutes. In some cases, travel times exceed 60 minutes even from very close districts (e.g., the district of Notre-Dame-aux-Neiges in the north-east of the city centre, between Botanique and Madou, less than 1,5 km as the crow flies). In addition, several districts are completely excluded due to the lack of accessible stops and/or vehicles for all the consecutive stops. The map also shows the detrimental effect of the underground network, which never satisfies the hard step-free constraint. Indeed, areas from which travellers would normally access the inner city through a combination of surface and underground services lose the benefit of the latter. Travellers then have to combine a number of surface lines, still under the hard step-free constraint. This explains the large gap in travel times compared to the no-constraint results in both eastern districts (tram lines 39/44) and the north-western communes west of the underground (Koekelberg, Berchem, Ganshoren and Jette).
40Moving on to the tactile paving constraint, the results are almost as worrying as for “hard step-free”, although here less stops are excluded (but then with very long travel times, usually over 75 minutes).
Figure 2. Travel time to Bourse
41Figure 3 presents the results for the Brussels-Midi rail station, which is a major regional, national and international transport hub that includes high-speed rail and coach services. The area is also home to a number of offices and a very large street market on Sundays.
- 10 It is worth noting that for those experiencing the longer journeys, most train services can also be (...)
- 11 Although any excluded district is of course regrettable.
42Travel times without constraint show a similar spatial pattern to that of the Bourse destination. The variety of PT routes and the different (pre)metro lines going in several directions make it possible to reach the station in less than 45 minutes from most districts across the Region10. As with the Bourse stop, the soft step-free constraint induces a moderate increase in travel times and a limited number of excluded districts11.
43Imposing the hard step-free constraint means a massive increase in travel times to Midi station, which is eventually even worse than the Bourse case. Travel times from almost all districts are over 60 minutes, and actually often over 90 minutes, including from central districts. This is because only one stop at Midi station (served by bus line 48 and located on a nearby street) meets the “hard step-free” condition. Virtually none of all the other PT stops and/or vehicles serving Midi station (namely, the underground lines and the surface tram and bus stops located under the railway bridge) satisfy the hard step-free constraint. As a result, the only way to reach Midi station under the “hard step-free” constraint is either to start from a stop on route 48 (provided it is inclusive in its design) or to combine one or more bus/tram lines (the underground lines being excluded by the constraint) until one reaches line 48. This is the price to pay for finally getting off at Midi station. In concrete terms, itineraries become more complex with more routes to be used, and thus a dramatic increase in the number of transfers faced by disabled travellers.
- 12 Obviously, it is extremely unlikely that a disabled person would undertake such a journey.
44This is evidenced by Figure 4, which illustrates the overall loss of direct access to Midi station. The case of the north-western district is particularly open to question, with no less than four transfers and thus five successive lines needed to reach Midi station12. This is due to a twofold effect. First, let us recall that the underground is not an option under the hard step-free constraint. Then, being forced to use the surface network, disabled travellers in need of the hard step-free option face short lines that have been designed to carry travellers to the underground instead of going beyond. This contrasts with other cities, such as Zurich or Vienna, which have historically favoured longer lines across the city to ensure that the probability of transfers is lower. In this debate between shorter and longer PT routes, it is worth noting that considering unconstrained accessibility, Lebrun [2018a: 169] found that a significant proportion of PT travel time in Brussels is linked to the number of transfers faced by passengers.
45Furthermore, the hard step-free constraint also excludes many districts, roughly the same as travelling to Bourse.
46Again, as in the Bourse case, travel times under the tactile paving constraint are somewhat shorter than under the hard step-free constraint, although dramatically higher than without constraint (Figure 3). However, the deterioration of travel times compared to no constraint is even worse than for the Bourse. A similar conclusion can be drawn for the number of transfers (Figure 4).
Figure 3. Travel time to Brussels-Midi rail station
Figure 4. Number of transfers to Brussels-Midi rail station
- 13 Of which one bus route is operated on an hourly basis only.
47Let us move on to ULB Solbosch, which is the main university campus in Brussels. Its location on a semi-peripheral site in the south/south-east and the fact that PT services are restricted to four surface routes13 explain that unconstrained travel times are usually higher than to the Bourse or Midi stations, which both have underground services and numerous surface lines.
48Again, imposing the soft step-free constraint does not make much difference and excludes the same stops as in previous case studies. In contrast, the hard step-free constraint leads to a dramatic increase in travel times. The spatial extent of the 90+ minutes area is even larger than for the Midi station. However, even when travel times are below 90 minutes, they are usually between 45 and 90 minutes, despite the proximity. The main reason for this is the inappropriate design of three of the five stops that serve the campus. As in previous case studies, itineraries to the campus then become so absurd that travel times soar (e.g. 52 minutes from Flagey Square instead of 13 minutes without constraint).
49Tactile-paving travel times are somewhat shorter than with the hard step-free constraint, but still much higher than when travelling without constraint. It would take more than an hour to reach the campus from roughly three quarters of the Region, and more than 90 minutes from about half of the departure districts, not to mention the usual excluded departure points.
Figure 5. Travel time to ULB Solbosch
50Finally, Figure 6 shows the results for UCL Saint-Luc, which is one of the three academic hospitals in Brussels and a higher-education campus. The site is located on the eastern edge of the Brussels-Capital Region. However, thanks to the underground, its unconstrained accessibility is good compared to other peripheral destinations. Bus routes 42 and 79 complement the underground service.
51Once again, imposing the soft step-free constraint does not make much difference, notwithstanding the same exclusion of departure stops as in previous case studies. In contrast, the hard step-free constraint leads to the worst increase in travel times among the four case studies in this paper, plus numerous excluded stops. Indeed, neither the underground station nor the bus 79’s stops satisfy the hard step-free constraint. Hard step-free access to the hospital is then entirely dependent on bus route 42, which provides a short service (4,8 km) to Roodebeek station. As with Midi station and ULB Solbosch, this results in indirect, odd routings, with many more transfers than without constraints (Figure 7). As in the case of trips to Midi station, the maps in Figure 7 show how much the non-inclusive nature of the Brussels’ underground creates a barrier between districts that normally count on underground lines to connect them, in the context of short surface lines that cannot provide a substitute for the metro. For instance, travelling from the western districts to Saint-Luc would require four transfers under the hard step-free constraint. Furthermore, the results for “tactile paving” are similar to those for “hard step-free”.
Figure 6. Travel time to UCL Saint-Luc
Figure 7. Number of transfers to UCL Saint-Luc
52Based mostly on open/free data and software, this paper is the first attempt to systematically quantify the gap in PT access to the city faced by disabled people in Brussels. To the best of our knowledge, similar investigations seem to be scarce, also in the context of other cities.
53At the time of our data (October 2022), travelling and dealing with constraints leads to suboptimal or non-optimal itineraries that often take more time (estimated by a travel modelling approach to +6 %~+68 % subject to travel constraints), require more transfers (+1 %~70 %) and involve more walking (+1 %~+54 %). Furthermore, the results are very heterogeneous, depending on the travel constraint, the departure point and the destination. Among the travel constraints considered in this paper, the most critical ones are hard step-free and tactile paving. It is worth emphasising that our results for these two travel constraints are likely conservative in the sense that in the real world, and according to our qualitative approach, many unforeseen events may prevent a disabled traveller from using PT, even if everything is designed appropriately. These include, for example, a lift or a wheelchair ramp that is out of order, a bus that does not stop close to the platform and illegal car parking at a PT stop.. In addition, the real state of the public spaces could not be considered.
- 14 The STIB-MIVB’s TaxiBus service is very useful and utilised by several disabled persons. However, i (...)
54We also found spectacular effects due to the non-inclusive design of the underground network. In an unconstrained context, the underground is the backbone of the STIB-MIVB network for both pull factors (efficiency, regularity) and push factors (network structure that forces users to combine surface lines and underground lines to access the city centre or to cross the city). The underground, however, becomes a real obstacle to PT travel under the hard step-free and tactile paving constraints. Its actual non-inclusive design can imply very illogical and painful surface routings. The underground thus plays opposite roles, depending on whether the individual is disabled or not. From this perspective, it is important that our model estimates itineraries whenever a theoretical solution is possible given the travel constraints under consideration. In the real world however, it is very unlikely that a disabled person would consider using PT travel when travel time is as long as 90 minutes or more and the number of transfers is three or even four. The alternatives would thus be a private car, a taxi adapted to specific needs or the STIB-MIVB’s TaxiBus service14. Otherwise, the journey would simply not be possible.
55All in all, public transport in Brussels is thus currently not truly public. Despite the significant efforts made in recent years, there is still a lot of room to make PT truly inclusive. A next step of the research will therefore be to investigate the benefits of various improvement strategies (e.g. is it more efficient to spread the improvement or to focus on key nodes?). The same methodology could also be replicated to assess subsequent improvements of accessibility across the STIB-MIVB network.
56It is worth noting that our analyses were based on the state of the infrastructure and the network (routes and timetables) in October 2022. Given regular upgrades of stops and rolling stocks as well as changes in the network design, it would make sense to regularly update our results. This could be done through an appropriate watchdog. The investigation could also be expanded to other days and times than an arrival at 9:00 on a weekday. Our investigation informs other cities with non-inclusive stops and/or vehicle design about potentially overlooked inequalities in accessibility. In addition, our analyses could be replicated to other cities for comparative purposes and to learn from the most inclusive cities.
57Beyond Brussels as a case study, this paper also demonstrates the power of quantitative investigations guided by a qualitative approach. We trust our results will add even more legitimacy to the demands of disabled people and NGOs defending their rights. It is unlikely that policymakers and public authorities as a whole are aware of the gap in PT access to the city we have found in this research. This paper should therefore help them to appreciate the need for rapid and massive improvements. In this context, we also hope that this paper will encourage more scholars from the field of disability studies to consider quantitative approaches.
We express our gratitude to all the disabled persons and partners’ employees who contributed to this research through interviews, meetings and trips across the city. We also took advantage of the very stimulating and useful comments received from Bruxelles Mobilité, STIB-MIVB and CAWaB on an earlier draft. These fruitful exchanges totally respected our academic freedom. Lastly, we used open-source software and data as part of this research and would like to thank all those who help keep open-source communities alive.