Step 1: Update database with the latest counts
Before starting the process, all available count data needs to have been loaded into the database. In RAMM, this can be achieved through either loading the results manually, using the MetroCount load facility or the bulk importing of count records. A number of audit checks can be done at this stage also.
Step 2: Update new sections/sections with no estimate
Checks should be made that each carriageway section has an estimate. These should be manually confirmed but there are various methods to establish an initial count such as looking at an average of the ONRC band the road is classified in to.
Secondly, consideration should be made of changes to the network and whether localised links or estimates should be updated. Changes in the network leading to step changes in traffic flows or new road infrastructure such as bypasses, attraction facilities or land use change should trigger review.
Step 3: Establishing Groups
Road types are collections of links (or sections) that have similar profiles for volume variations through the year, due to seasonal variations. The use of seasonal variation profiles is very much at the discretion of the asset manager. As such, the default type would be a single road type with a default adjustment factor of one. There can be then increasing levels of complexity and granularity as the asset manager chooses.
Traffic groups are collections of links (or sections) that have similar profiles for typically land use or ONRC class. Their purpose is to utilise smaller sample sizes to address links with no count history to update.
It is recommended that traffic groups are maintained. It would essentially be a manual process to generate these initially. There are options to automate the task, albeit outside any proposed RAMM tool. Any process for traffic groups within RAMM would be at a very high level and not reach the benefits of a realistic study of traffic mix. Grouping is a good method of covering vehicle mix and similar performing links. It is therefore encouraged as part of a recommended practice. It is accepted assuming consistent growth within a group is likely not realistic. However, grouping does serve other purposes such as traffic mix. Grouping can be adopted to the level the user chooses including just having one!
Step 4: Estimating sections with counts since the last estimate
Feedback showed various ways of doing this from assuming the last count to seasonal/daily adjustment factors across the industry. The guideline recommends a reasonable minimum standard. From here, others can apply more complexity if they wish and some methodologies are suggested and discussed. This may depend on the complexity or characteristics of the network or parts of it. It is noted that some roads do have significant seasonal variations. For example, care should be taken where seal designs for summer traffic volumes are based on estimates from a winter count.
The minimum standard would be to simply update estimates based on the latest count. A more complex method proposed is a mix of automation via sql and audit checks. Where there are at least three counts over the last six years from the same location within the link, a trend analysis can be used to determine the estimate. The minimum three counts, including the latest count, must be within say greater of 20% or 50 vpd of each other. This method will suit higher volume annual and bi-annual sites with a regular count history.
Where sites do not have a regular consistent count history, the latest count will be the basis for updating the estimate, should it meet auditable criteria such as being within the maximum of say 30% of the previous estimate or 50 vpd or more sophisticated to vary by ONRC category. (These criteria would need to be agreed as the previous estimate could be misleading, rather than the count.). Counts outside this variation would be flagged for a manual check or confirmation.
The estimate is dated to the year of the last count. The estimate can then be updated to the current year by utilising the annual growth group prediction.
It is recommended that any automated process include some measure to distinguish between updating links which have had a count and a full annual update across the network. It is recommended that only one estimate is allowed in a calendar year as an example and any are overwritten each time within that period. There should always be the ability to manually update any estimates based on local knowledge.
Step 5: Estimating sections with adjacent counts since the last estimate
Typically, a mix of automated to start with but equally sanity checks with manual checks to see if makes sense. Determining estimates pro-rated for carriageway sections between links with a count-based estimate by SQL with manual checks along routes to adjust any anomalies. This is recommended, again with an audit step like that for links with counts. Utilising adjacent data is better than treating as a link with no count data.
Step 6: Estimating sections with no recent or adjacent counts
A recommended approach is using growth groups and assigning an average growth adjustment on those links left in the grouping not adjusted directly in the previous steps. This can then be over-ridden/adjusted by a manual adjustment. This could allow for groups of stable areas or no exit roads being assigned to a “fixed estimate” traffic group where the links are given a zero growth parameter .
Generally, higher classification routes will have more regular counts being undertaken and so will typically have more information to draw from. Lower classification links can be more variable from well-established areas with little or no change to development areas with significant growth. They typically have less frequent counts and more reliance on estimates. The assignment of groups can mean those fast-changing areas are identified and separated out. This can then be addressed by assigning more frequent counts at these sites than other similar classification groups in low growth areas. Secondly, we can assign more attention in manual checks on these groups to understand changes and adjust estimates as needed, using trip generation factors for example.
Step 7: Traffic Mix
The issue is around moving from default assessments to more accurate estimates for traffic loading. It is recommended to adopt a similar approach in volume estimates i.e. directly updating links associated on or near count sites and then adjustments by averaging mix over the remaining links in the group. The count mix can be used if six or less years ago. Otherwise use an average of count valid six years old or less from the same road. If none apply, then use an average from the road group.
The analysis is recommended in conjunction with the traffic estimate process. The ability to manual adjustments should be included.
Step 8: Auditing
At the end of the process, audit checks are undertaken to assess estimate confidence and approve estimates for entry into the database.
Step 9: Links
The use of links does help. Links reduce the number of similar performing sections, improving and focussing the sampling analysis for completing a traffic count site strategy and programme. The Link process is already established within the RAMM Traffic Count Estimation (TCE) tool. Some recommendations as to considering whether a user-defined threshold is needed. A recommended improvement is having a minimum value, so the threshold is a maximum of say 10% or 50 vpd. It should also allow for manual adjustment of links due to audit reviews and quality checks. Links should be reviewed every 3-5 years.
Step 10: Determine Count Strategy
Generally, in place already but if starting from scratch this would sit here in the order. This should be established in accordance with the existing best practice guidelines around rotational and core sites.