Requirements for Freight Modelling �Now and in the Future

Ian Williams, Independent transport consultant

57th Transport Modelling Forum

Birmingham, 9 July 2026

Intro: Passenger Modelling v Freight Modelling

• Passenger demand models have been widespread across GB
• Freight demand models have not
• LGV models are even rarer
• All these 3 types of transport are important to support the national economy and lifestyle
• TAG 2.1 Variable Demand Modelling states w.r.t. passenger trip purposes
• “A distinction between purposes is however essential … A suitable starting point would be - �commuting, employer's business and others” (2.6.5)
• “Not all stages of the demand model require the same degree of segmentation” (2.6.6)
• People movements are more straightforward to predict than goods movements
• Freight transport demand patterns are much more heterogeneous than passengers’
• Commodity type (NST: 20 categories)
• Intermodal flows (Container ship - Shanghai to Felixstowe, rail to Midlands rail terminal, HGV to NDC)
• Supply chains, distribution channels and logistic legs
• Freight models need to employ a realistic behavioural-economic foundation
• That explicitly represents the cost structures, that govern the choices that underpin goods movement patterns
• Why goods are moved - is as important as where (OD matrix), what (commodity type), and how (mode)

​

​

Presentation Summary

• Evidence on freight movement patterns and heterogeneity

​

• Logistic systems, distribution channels and logistic legs

​

• Requirements for modelling freight at present

​

• Future modelling challenges - as transport technologies evolve

Understanding freight movement patterns

• If you cannot explain the current movement pattern of freight
• Why should you believe that you can forecast how it would respond to future policy initiatives?
• The types of requirements from a mode/vehicle type are systematically different between:
• Different types of goods
• expensive, fragile, perishable, heavy, bulk solid/liquid, etc.
• For any specific type of good - between different legs within its sequence of logistic and intermodal legs from producer to consumer
• Non-urgent legs - to national distribution centre (NDC) or port bulk silo
• Urgent legs - parts to a repair firm - accompanied HGV to daily RORO ferry
• Consignment size matters
• moving 1 ton every few weeks versus moving 100 or 1000 tonnes every day?

Evidence on road freight vehicle movement patterns and heterogeneity�

Ratio of Artics to Rigids - Motorways v Urban A Roads

The intensity of road usage for Rigids v Artics varies by Road type by Time of day, across the Week

Mix of HGV sizes varies greatly by context

TIME PROFILE TRAFFIC INDICES FOR EACH OF ARTIC AND RIGID, BY DAY OF WEEK ON MOTORWAYS, 2018, GB

For Artics on motorways mid week, the 1 AM traffic is 1/3 the peak midday. Artics avoid Saturday night to 4 AM Monday

For Rigids on motorways mid week, the 1 AM traffic is 1/9 the peak midday. Rigids concentrate: 6 AM to 4 PM weekdays

Different sizes of Rigids have different usage patterns by road type and time of day / week.

​

Veh. kilometres by HGV type - % by road type, 2024 GB

(Source: DfT Traffic Table TRA301)

• Vehicle kilometres of the largest artic are focused away from urban (4.7%) and minor roads (1.6%)�but onto motorways (59.7%)
• 2 and 3 axle rigids have a less skewed pattern of vkms across road types
• HGV traffic over time is becoming more concentrated onto motorways, away from urban and minor roads

Trips by HGV type - % by length of haul: 2024 GB

(Source: DfT CSRGT Table RFS0113)

• Length of haul of the larger artics is focused away from short trips�but onto trips of 100 kms or more
• Smaller rigids have the reverse pattern of length of haul distances
• But Largest Rigids >25 tonnes have the lowest % of trips >100 kms
• The COBA/TAG OGV2 class that mixes artics and large rigids - is not homogeneous in trip patterns!

Average load per loaded vehicle by commodity & HGV type

Source: CSRGT data from 2004-07 for GB, processed for BYFM model

NST:4 – Food products, beverages and tobacco

• Artic. average load is 6 times that of rigids<17t

​

NST:11 – Machinery & Equipment

• Artic. average load is 2 times that of rigids<17t

​

Average load per vehicle by commodity type

• Varies little between most commodity types for small rigids
• Varies by a factor of 2 or 3 for artics

​

Models need to take account of

• Vehicle type - not simply OGV1 & OGV2
• Commodity type

Logistic systems, distribution channels and logistic legs�

Cost effective parcel delivery from a container port to my home – why logistics matters

Production-consumption, distribution chain and distribution channels

View of: the public the logistics industry the modeller

Logit discrete choice model between� competing distribution channels

Logistic systems, distribution channels and logistic legs�

• Logistic systems are designed to minimise the total cost of distribution of goods, through balancing
• Minimising the distance travelled by the goods
• Minimising the vehicle operating costs/tonne carried
• Maximising consignment sizes
• Minimising warehousing and storage costs
• While avoiding stock-outs
• The optimal logistic structure and its component distribution channels will depend at any one time on
• The warehousing technology and operating cost structures of the distribution centres
• The transport technology and operating cost structures of the vehicles and modes
• It is typically faster and cheaper to build new distribution sheds of optimal size and location
• Than to relocate productive industry sites
• Even if the Production – Consumption (PA) matrix of trade of goods is little changed over time
• The set of individual goods trip legs is likely to change
• Because distribution channels adjust to cost changes

​

Example of distribution chain and logistic legs for consumer goods

Logistics structures – Evolve quickly

• Growth of Port-Centric logistics
• In mid-2025 Tesco confirmed plans for a major investment in a new distribution hub, due to open in 2029, at DP World’s London Gateway port.
• Tesco states that this investment is to ensure its distribution network remains fit for the future.

​

• Growth of fast fashion
• Parcels from China delivered direct to UK consumers
• Brands bypass local inventories across Europe by buying up spare, discounted airline space and using direct parcel drops. Air freight 3-7 days, express courier 2 to 5 days.

​

• Dark warehouses and fulfilment centres –Amazon, Ocado
• Automation of picking by robots over 24 hours, not by humans over 8 hours
• Less workforce needed
• Lower warehousing cost per consignment

Requirements for modelling freight -�at present�

Main UK freight related data sources

The commodity/product/industry type categorisation adopted in its statistics is generally mode/sector specific

As is the weight definition for a consignment/container …

Inconsistency across data sources is the norm – Each was designed to meet the specific needs of its own sector

BYFM freight traffic is estimated in 6 stages�

Goods, services and overseas trade

in the UK economy

Supplied to firms, institutions, shops & exporters

to meet their needs

​

Through distribution stages

(Producers – logistic chain – consumers)

Generating movements of goods

(and handling at each distribution stage)

Choice between road and rail

Then assignment of traffic on networks

Freight

demand

Transport & logistic

costs

BYFM structure – Freight demand/supply

Structure of TRIMODE model

TRIMODE has a modular structure

21

PASSENGER DEMAND MODEL

ASSIGNMENT MODEL

TRANSPORT MODEL

FREIGHT DEMAND MODEL

ECONOMY MODEL

NATIONAL MODEL

REGIONALISATION MODEL

ENERGY MODEL

FLEET MODELS

ENERGY MODEL

Transport activity

Operating costs

Vehicle costs

Transport energy consumption

Accessibility

Trade growth

Income growth

Income growth

Freight demand modules

Future modelling challenges – �as transport technologies evolve�

Evolving major transport supply changes

• Electrified large HGVs – 29% of sales in China in first 4 months of 2026 were BEV
• Sustainability benefits from rail may dissipate significantly
• Autonomous HGVs – these are most likely to be feasible on motorways for long haul primary logistics – already in use in parts of the USA
• Removes truck driver costs and allows 24-hour vehicle usage (no driver rest time needed)
• Radically improves truck cost competitiveness relative to rail
• Dark warehouses
• Radically alter the cost functions and technology of supply chains
• potentially lead to major changes in the geography and structure of distribution legs
• I am old enough to remember the disappearance in the 1970s and 80s of:
• The dock workers, the coal miners, the print operatives, etc.
• So - when there are major technological revolutions already underway in transport
• We need sound behavioural-economic model structures to envisage both:
• their direct transport impacts
• and their potential wider economic impacts

Thank you – ��Questions?�