Maaike Snelder has done interesting research on the robustness of a road network and tested how robust the Dutch road network is. A well designed main and secundary road network contributes to robustness. To find the thesis google: "Designing Robust Road Networks" - TU Delft Institutional Repository

In my opinion good idea is to make some kind of road inspection. For example if road fulfil standards it should have according to hierarchy. In Poland we have such a problem, that lots of roads are assigned to road category "x" but its real state suggest lower category...

Maybe by one topological approach...

Please check this article: "User-Oriented model to support funding decisions in pavement management" by Mellano, Sassanelli and myself. We present a model to set up a hierarchy of interventions on road networks. Additionally, an article is forthcoming on the Journal of Computer-Aided Civil and Infrastructure Engineering, proposing a new index of road quality, which incorporates LoS and IRI

The US highway system is complex with at least three parameters to consider. The first is ownership which includes but is not limited to Federal, State, County, Local and toll road usually owned by an authority: Semi-public) The second parameter is Functional Classification(its operational characteristics) including arterial, major and minor collectors and local. The third is funding source-Federal Aid being the biggest. In addition there exists the NHS.....National Highway System which includes the 1956 expressways constructed as the Interstate System and its subsequent additional segments.In essence, the US highway network goes back to the 1916 Federal Road Act that provided substantial government funding to the states with the proviso that each state initiate a State Highway Department through which government funding would be channeled and standards established for its use. Over time the Primary Federal Aid system and the Secondary Federal Aid System were created. State Highway Departments currently have sophisticated software to "model" and test a wide array of typical and atypical highway design features arriving at a solution that provides an efficient and effective highway system.

G. Larry Larned

Thanks very much to everyone for the responses.

Amelia - when you say road assessment for services, do you mean for example level of service or congestion? What I had in mind was: Someone assigns a functional road classification to a network. How do I check it is correct? Perhaps to look at connectivity, or perhaps to somehow see if regional distributors form a network, or perhaps - if a regional distributor is defined as carrying high speed, long distance traffic then to measure % vehicle trips against journey distance.

Atze, thanks for the document, I will read it and get back.

Diana, thanks for the tip, I will find and read the thesis and get back to you.

Joanna, you are right, I mentioned this in my blog post on road classification in Poland, see link below.

Gildemir, could you give an example of what you mean?

Mauro, thanks for the tip, I will google the document. If you have other material in Italian, I could read the original document.

Larry, I need to think about what you describe. Perhaps it covers "classification" of a road network more than "assessment"? I also would be interested in what state highway departments use - maybe there is a use for GIS in testing a road hierarchy.

Adolfo, I believe road classification can cover traffic volumes, and for road hierarchy traffic volumes are sometimes assessed in terms of level of service. You are right about road hierarchy and vehicle activity in urban areas. There was a European research programme a few years ago called ARTISTS which looked at this in detail. The problem then is also that a road could be a link in several networks, and a different hierarchy in each. For example, a shopping street could be a district distributor in a vehicle road hierarchy and a local access in the pedestrian road hierarchy. Maybe that is where things begin to go wrong.

http://wp.me/p27Pox-3U

I refer you HCM(HIGHWAY CAPACIY MANUAL) , AASHTO guide or GREEN book.

You must determine road classification based on road capacity,the number of lines and acccess points,mean speed,density,lane width and other parameters that you study within above refrences.

and Traffic Engineering from Mcshane

About Topology as a criteria (Gildemir's suggestion) - you could determine the importance of a road segment by how often it is used in routes between different origins and destinations.

An algorithm to determine this would overlay a map with a grid of possible start or end nodes for routings and calculate shortest/fastest paths between each pair of nodes. A measure of connectivity for road segments would be how often they are ocurring in these paths. For each routing additional informations (e.g. population density at origin/destination) could be taken into account.

May be by getting help from artificial intelligence using advance unsupervised clustering method and compare the result by what it is now known as hierarchy of roads and see how much does it works right feel free to get more information about this advance methods from me any time.

Thanks Günter, your idea sounds fine; it could be done with a GIS, and with added features such as "what % of trips on a higher level of road are actually long distance trips" and "what % of a long distance trip is made on a higher level road".

Ali, you are ahead of me - could you suggest a text which would give me an introduction to your suggested topic please?

Robert, sorry I'm entering this very interesting discussion now. Please I need some clarifications. Can you define the words 'right' and 'works'. Are you envisaging the workability of a road network by a variable or several variables? If so, what are they?

Robert, to some extent visual surveys with trained panel members and the functional surveys using technologies like ROMDAS (ROad Measurement and Data Acquisition System) might help in assessing the road networks.

The question to ask is "Will the average nationwide trip time be lower after your proposed interventions?" To answer this, you first need detailled data on origin-destination traffic volumes, link capacities and actual routings. You cannot work in a theoretical vacuum.

Thanks everyone for the good comments. Please give me a day or so to reply in more detail.

No need to reply. Just do the research!

Hi Robert, I agree with Ali's comment, and would like to specify that you could use the Analytic Hierarchy Process (AHP), to get a new multicriteria hierarchy of roads. There are examples in MATLAB even for fuzzy AHP, which I consider really interesting. See http://www.mathworks.it/matlabcentral/fileexchange/33406-fuzzy-ahp/content/FuzzyAHP.m

Hallo Christie, fair comment. My original question really should have been, what is the purpose of a road network, and how do we measure if a road network is meeting this purpose? Since I posted my original question I have discovered that there are many ways in which people "classify" roads (see my publication). Then, partly through suggestions in previous comments, I found ideas for what could be useful indicators of "purpose" - for example: connectivity, accessibility, route factor, consistency, robustness. It begins to look like road network analysis could be a new discipline all by itself.

Praveen, yes - even the development of a structured road network can be done using people rather than computers. One research paper talked about a methodology using a workshop situation, where the "stakeholders" themselves participate in the network design process. By the way, I need to read up about ROMDAS before I can sensibly comment on it.

Ian, maybe average trip times are not a good indicator of whether a road network is right or not. For example there is the link between land use and road networks (If you change one you change the other). Any increased travel times could be due to poor land use planning, so that a better solution to building more roads would be to correct the land use planning. It would also be nice to avoid the use of complex traffic models, since these can hide assumptions made in the analysis. Simpler tools, like tests for connectivity or accessibility would give a more open, practical approach.

Mauro, I need to brush up my understanding of AHP and fuzzy analysis first. Your ideas on a user-oriented index are interesting. You mention elsewhere an index which had more than 50 indicators, and I have found at least 50 ways in which people" classify" roads.

To add another angle, I wonder if has anyone used 3D-GIS to view and illustrate multi-variable presentations of a highway network? Indeed I think that general use of a GIS to study a spatial object such as a road network (and features such as connectivity, accessibility) could make life easier for the analyst and his audience.

Data GTA-304: A List of Road Classifications

Hi Robert,

we also used a Fuzzy Logic Approach for ranking transport infrastructures

Please see the publication below or the attached one. BR

https://www.researchgate.net/publication/225931588_A_Fuzzy_Logic-Based_Methodology_for_Ranking_Transport_Infrastructures?ev=srch_pub

Chapter A Fuzzy Logic-Based Methodology for Ranking Transport Infrastructures

I hesitate to jump in at this late stage, however, you may find the OECD publication, "Safety Strategies For Rural Roads" published by the OECD in 1999 very helpful. The book is now, as far as I know, only available as a paid-for PDF download from the OECD or I borrowed it for a while, before purchasing my own copy, from the British Library.

I was looking at it from a road safety perspective and the book points out that the wide variety of classification schemes in use makes it hard to compare road safety between different countries and in fact obscures the problem. One problem I think business contributes is a desire to have ever faster movement of traffic without necessarily considering road safety. The 3 road classifications I find most helpful in looking at road safety are "Strategic Routes", these are the long distance communication roads between different ends of the country, such as motorways and good class "A" or trunk roads; "Distributor Roads", which are the roads that take traffic around a particular area and "Access Roads", which are purely for accessing various properties. In terms of road safety, each of these types of road will have different types of traffic and speeds and it is best not to allow roads to have more than one of these functions, otherwise you get more road traffic collisions. For instance, motorways have very few access points, bicycles, pedestrians and slow moving vehicles are banned and their speed is high, for efficient transport. At the other end of the scale, access roads may well have mixed traffic types, including pedestrians and bicycles, as well as larger motorised vehicles but their speed should be lower and vehicles should not be using them as part of their long distance travel needs.

Thanks Robert for your response to my comments. I totally agree with you that Road Network Analysis could be a discipline all by itself! A few years ago I did a research on the road network of a Study area based on connectivity and accessibility. It was intriguing. A minute part of any road network analysis spans across all disciplines and human activities.

You are welcome, Christy - and I think GIS would be a good tool for network studies (although that is another topic).

Michele, thanks very much for the links - I have downloaded your two documents on fuzzy logic approach and hope to read through them this weekend.

Hallo Gwyneth, Thanks for your detailed comments. I found the document you referred to on the OECD website (it is possible to read it online, but not to download it). I would agree with the document about the large variety of road classification schemes in use, but I think the concept of "functional classification" (which might be a bit like your 3 road classes) is past its sell-by date. A simple, globally applied classification would be a good idea. Separating traffic in terms of speed potential or type (cars / bicycles / pedestrians) is ok for some roads but not for others (as you say). I think the cut-off point is maybe a design speed of 70 km/hr. Not sure about the value of long-distance routes per se. Some countries (such as Germany) plan their structured road network by basing it on a structured land use (which is as it should be). This offers a way of planning and testing a road network with less consideration for "long distances". Finally, the emphasis you put on safety is right. I need to read again some earlier documents on road network testing in terms of safety - one problem could be engineers thinking in terms of individual link design, Y

Sorry Gwyneth (I pressed the wrong button). You might find this weblink interesting.

Regards, Robert

http://wiki.coe.neu.edu/groups/nl2011transpo/wiki/08749/

It starts with having a good travel forecasting model to determine where people want to go and why they want to go there, on both a microscopic and macroscopic level. Use this model to load your network to determine whether it is functioning as it is supposed to function. If you don't have modeling capability, you can do it through field observation. Realistically, you should end up with two different networks- a theoretical planned network, and an actual operational one. The final step is to synthesize the two networks. This can be accomplished in one of two ways- changes to the operations of the network to get it to perform as planned, or changes to the planned network and land uses to better match the reality of how things are.

Poor classification usually means problems. You might look at traffic safety, noise etc. When hierarchy of technical parameters are incoherent, accidents and law abuse happen more often.

classification is usually defined by government hence it also difficult to change untill taking them into stake.

Hallo Ravindra. What you wrote made me think. Certainly government has some involvement, but the extent probably differs depending whether you ask an engineer or a government member. However maybe what I should have asked was "how can we define a "right" road hierarchy?" (on the idea that if we have a definition then we can measure the hierarchy against it). In fact, maybe I should have asked about a "right" road network. One definition of this might then be "the network which most reduces travel time for shorter trips made by traffic modes which are environmentally friendly" (so not favourable to longer distance or to car trips). Such definitions would be understandable by decision-makers and probably also measurable.

by serving the intended objective 

I agree about the classification type.

Speed limit and other parameter like traffic is important for hierarchy. 

What about socio-economic perspectives to the hierarchy?

Road hierarchy is completely different than social hierarchy. Traffic flow, road width and its restrains( parking, stops etc) in urban area is major parameters that define Road hierarchy

https://dot.sd.gov/media/documents/HwyFunctionalClassification.pdf