Thursday, 28 December 2017

Does ordering stuff online increase carbon emissions?

How much of your shopping do you do online? Are you worried about the climate change impacts of online shopping? A recent study of parcel delivery in Central London [1] gives some insights into how this compares with shopping on the high street. London is a bit different from the rest of the country: I have extrapolated these results to Cambridge, with plausible assumptions.

Delivery in Central London is equivalent to driving about 1km in a car, for Cambridge it is nearer 3km. If the parcel is plastic wrapped the carbon emissions for packaging are small but for a medium sized box you need to add another 1.3 km. There are a number of ways in which these emissions could be reduced.

A third of all parcels are clothing and footwear; 20-30% of this is returned.
As of 2016 online shopping accounted for 14% of all retail spending in the UK. The chart below shows the different kinds of items we have delivered. Only 8% of these items fit through the letter box so the delivery man has to wait for someone to answer the door. A large proportion is clothing and 20-30% of this is returned, which means another visit to collect the unwanted goods.

Data from [1]

All these parcels add to the traffic on our roads. Light commercial vehicles now account for 15% of all traffic (by distance travelled) and this is up by 52% since 2000. Over the same period car travel has increased only 8% [2].

When you buy from a shop, the goods have come from a supplier warehouse to the shop; when you buy online you leave out the shop but you add at least one warehouse, which is the local delivery depot. Arguably the journey of your purchase to a shop is comparable to the journey of your online shopping item to the depot. In this analysis I consider only the journey from the local delivery depot, comparing this with how you might have travelled to a shop.

The study: average 72 customers in 7.3 hours, 11.9km driven
Researchers accompanied delivery vehicles on a sample of 25 trips in the Westminster and Camden areas delivering non-food parcels. On average, trips took 7.3 hours and visited 72 customers handling (delivering or collecting) 127 parcels. The driving distance within the delivery area was 11.9 km [1].

Perhaps you won’t be surprised that the vehicle was parked just over 60% of the time and the driver walked another 7.9km to and from each home. The focus of the research was the cost of deliveries made increasingly difficult by restrictions on driving and parking and general congestion. The average driving speed in the delivery area was just 7 km/hr which seems astonishingly low. This must be partly due to congestion but the average speed in the congestion charging zone is double the speed these vans managed [1]. Another factor will be the driver going slowly to check house numbers or to find a parking space.

Stem distance from the depot to the first call: another 18km in Paris/London, 65km for Cambridge
The study did not report on the ‘stem’ distance from the parcel distribution centre to the delivery area. Perhaps this information is too commercially sensitive. They did mention stem distances found in other studies, for example in Paris this is typically 18km each way. Per customer that would be 250m. I have assumed London is similar.

Cambridge is rather different from Central London. Our parcels are usually delivered from Peterborough which is 65 km away. Also, the population density of Cambridge is about a quarter that of Central London, so the distance travelled between customers is double. On the other hand, the congestion in Cambridge is not as bad as in London (usually) so I have used the standard emissions/km for the van; in the Central London case I added a factor for congestion during the round, though not the stem.

A cardboard box could be 15 times the emissions from a plastic bag, but it is bio-degradable.
Online shopping often means more packaging than if you bought it in a shop. This could mean an extra plastic bag or a cardboard box. In terms of carbon emissions, a cardboard box is much worse than plastic. In the chart below I have allowed for a medium sized cardboard box weighting 250g. A typical plastic bag would be only 1/15th the GHG emissions, but some items need a box. At least cardboard is bio-degradable.

GHG emissions for delivery in Central London compared with Cambridge. Stem is van emission from the depot to the first call and from the last call back to the depot. Round is van emissions inbetween. For data sources see Detailed Assumptions below.

Equivalent distance in a car: 1.0 km (London)/2.7 km (Cambridge) plus 1.2 km for a box.

Suppose you have the opportunity to shop by car instead of online. The online shopping would be less carbon emissions if the car trip was at least 1.0 km in central London or 2.7 km in Cambridge (counting there and back). The box adds another 1.2 km. If you would have walked or cycled to the shop then online is always worse.

The emissions from making a pair of jeans are 20 times the delivery.
It is also worth comparing the emissions from the shopping journey to that of the package contents, but this is also highly variable. A 300g book would only need a small box but the book itself would be about the same as a large one. A DVD would be a little more. However for clothing, the emissions of the product dwarf the journey: for a pair of jeans in a plastic bag the emissions would be at least 20 times the parcel delivery.

How can we reduce these emissions? In the Central London study the researchers were focussed on reducing costs but some of their suggestions reduce emissions too. For example:

Click and collect services, or local collection points allow lots of parcels to be delivered to the same address. This reduces emissions as long as you would have visited the collection point anyway, or if you don’t drive further to the collection point than the van would drive to you. It also reduces costs and emissions due to failure to deliver: the vast majority of parcels will not go through a standard sized letter box and if the householder is out and a neighbour can’t be found then the parcel will have to be taken back to the distribution centre and delivered again.

Collaboration between parcel carriers would allow better optimisation of rounds, but this would have little impact in Cambridge because the distance to the depot is much greater than the distance between customers.

Bringing the depot closer to where people live would make a big difference in Cambridge. One approach is for the city to provide logistic services to the parcel carriers - this is being tried in Paris where they call it a logistics hotel.

Crowd shipping is an approach to reduce extra miles of travel using vehicles (private cars, taxis or delivery vans) who are going that way anyway and have spare capacity. There are several online platforms in the US for this such as Postmates and Zipments.

Even with zero emissions vehicles, there is still the problem of the cardboard box.
Zero emissions delivery vehicles would reduce the GHG emissions from the van to nothing, but there is still the problem of the extra packaging. Recycling the cardboard does not help very much because this still means breaking it down into fibres and remaking the card. Unfortunately it is hard to see how to make durable boxes that can be collected and reused without adding significantly to the costs.

Detailed assumptions:

I have assumed that the emissions up to the parcel dispatch centre are equivalent to the emissions up to delivery to the shop. In practice this is highly uncertain and variable. Freight to the shop will be more efficient because the goods will be packed more densely, for example in crates, probably in larger lorries, and with less packaging, at least for some types of goods. However, the distance travelled to the shop could be more or less than to the parcel centre, so this is hard to judge.

I have also ignored the energy consumed in the shop itself. Shops consume about four times as much energy as warehouses for the same floor area. However, warehouses tend to be large and we have a lot of them so overall shops account for less than twice the energy use of warehouses [3]. Also, the space occupied by shops depends more on the number of different types of goods than the volume of goods sold. High volume stores will have more stock space but this is not heated and lit to the same degree as the rest of the shop - it is effectively a small warehouse. I assumed that the marginal increase in energy consumption by shops for each item sold is small.

For Central London, the round distance is 11.9 km visiting 72 households, as per the study. Stem distance is 18km based on Paris. Emissions factors are as for a Class II van (1.3-1.7 tonnes) [4] times 1.6 to allow for congestion.

The population density of Central London is around 11,500/km2. The population density of Cambridge is on average around 3,200/km2 – so a quarter [5]. That means distances travelled in Cambridge per customers would be about double that in central London.

For Cambridge, visiting the same number of households, the round distance is as for London times two for the population density. The stem distance is 65km each way, to Peterborough. The van is the same. The number of visits is quite important in this calculation as the stem distance is shared between them. The longer distance takes extra drive time but less congestion and easier parking saves time on walking, so it is plausible that the same number of customers can be served in a similar time on the round. Also, there is a strong drive to reduce costs by filling the van and visiting as many customers as possible even if this means a longer duration trip which is a strain on the driver. I have therefore assumed there is no change in the number of visits per round.

A Class II van (1.3 to 1,64 tonnes) emissions are 0.23 kg CO2e/km
Cf medium car 0.18 kg [4]

Cardboard 0.84 kg CO2e/kg (0.79 if recycled) [4]
Plus disposal 0.022 kg/kg
Total 0.86 kg CO2e/kg
Medium sized box 300g so 0.26 kg CO2e/box

Plastic 2.6 kg CO2e/kg from [4]
Bag mass 5.5g so 0.0143 kg CO2e/bag

Book 0.3kg at 0.928 kg CO2e/kg [4]
Or, 1kg from [6]

Pair of jeans: 0.6 kg clothing at 22 kgCO2e/kg [2], total 13 kg.
Or, 6 kg from [6]

DVD 0.46 kg [7]

[1] Understanding the impact of e-commerce on last-mile light goods vehicle activity in urban areas: the case of London
Allen, J., Piecyk, M., Piotrowska. M., McLeod, F, Cherrett, T., Nguyen, T., Bektas, T., Bates, O., Friday, A., Wise, S. and Austwick, M
Transportation Research Part D Transport and Environment · July 2017
[2] Transport statistics Great Britain 2017 (
[3] The non-domestic National Energy Efficiency Data-Framework (ND-NEED) ( 2014
[4] Government emission conversion factors for greenhouse gas company reporting ( 2017

[5] Population densities from
Westminster 11,500
Camden 11,300

[6] How bad are Bananas by Mike Berners Lee (2010)
[7] Mobile network energy in fridge units (this blog) 2014

No comments:

Post a Comment