Successfully introducing a new fuel requires the coordination of many moving parts: Car manufacturers need to have vehicles ready for purchase. End-to-end logistics and technology infrastructure must be in place to deliver the fuel to consumers. The fuel itself needs to be readily and consistently available. And consumer demand has to be sufficient to ‘pull’ the product through this complex supply chain. Like any chain, this carefully choreographed organizational structure has inherent complexity and will break at the weakest point. Reducing the possibility of failure is an imperative in any system, but this is especially true with immature technologies operating in unknown market conditions. Facilitating an infrastructural medium is the key for consumer adaption for the new technology. Through comprehensive and ground-breaking analysis, Kalibrate was able to help ensure that hydrogen refuelling locations could be built where demand is most likely to exist.
A key part of the Department of Energy’s Hydrogen and Fuel Cells 2015 review was to analyse California data to ascertain where hydrogen was most likely to be in demand. Kalibrate worked with the National Renewable Energy Laboratory (NREL) in Golden, Colorado, to determine the factors that would drive demand for hydrogen and maximize the likelihood of it being successfully adopted as an alternative fuel in California.
This task, shaping the infrastructure of an emerging alternative fuel, was not a straightforward piece of analysis (and that’s an understatement). At the outset, Kalibrate knew that there is no data – anywhere in the world – that can be modelled to help understand what factors drive consumers to buy hydrogen. So the starting point of Kalibrate’s Research team’s work was to determine a proxy for hydrogen demand. The team tested numerous alternatives, and the most appropriate was determined to be gasoline expenditure. In technical terms, this became the Dependent Variable, and the focus then shifted to finding the factors that explained changes in that variable. If we could model that with any degree of confidence, then our recommendations to NREL and the DoE would be robust and reliable.
We assessed close to 31,000 locations using a combination of data provided by NREL, state and local government, third party data providers (Nielsen, RL Polk and Info USA data was used to compile select variables), gas station data collected via onsite visits by Kalibrate and modelling carried out by Kalibrate. We eliminated data having a low correlation to the dependent variable and also locations with less than 10,000 cars passing each day.
We then set to work assessing what could explain our Dependent Variable. Dozens of factors were considered and many eliminated. We eventually determined that there were 10 factors that explained our Dependent Variable and, therefore, could predict whether the location of a hydrogen refuelling station would be successful. They were:
|Households with income > $100,000||7.20|
|No. of gas stations||6.74|
|No. of hydrogen vehicles||6.67|
|Total no. of employees||6.09|
|Hydrogen permitting constraints||5.55|
|Gas stations with sufficient lot size||4.08|
|Hydrogen stations – existing/planned||2.32|
|Households with solar panels||2.31|
|Average commute time||1.01|
|Distance to hydrogen auto dealership||-2.40|
These factors explained the change in our Dependent Variable to an extremely high degree – an R2 of 0.89 was calculated. For the non-statisticians, this is the sort of outcome that results in academics and researchers opening bottles of champagne. We felt very confident that we could explain, with a very high degree of confidence, what would make a great location for hydrogen refuelling.
The impact for the retailer was clear: look for locations where these factors exist. Fuel retailers understand that each location is unique, and maximizing return on every location requires granular focus. That truism persists for hydrogen fuel. Ignoring the factors that determine demand will significantly impact a fuel retailer’s chance of success and potentially reduce the likelihood of hydrogen being purchased. There was, of course, a wider effect on the uptake of hydrogen as a fuel. If these factors were ignored and hydrogen refuelling locations built without considering these factors, consumers may become frustrated by the lack of availability of hydrogen. That impact could be significant and far-reaching.
Kalibrate used this modelling to determine spots across California where hydrogen refuelling locations would most likely be a success. We ranked about 31,000 locations across California to determine the very best locations in the state to build hydrogen refuelling sites.
Figure 1. Best locations for hydrogen refuelling in San Francisco
Figure 2. Best locations for hydrogen refuelling in Los Angeles
Finally, we predicted the number of refuelling stations that would be needed to meet the expected demand for hydrogen in California by 2025. We felt that a further 81 sites to augment the 11 that were built at the time and the 42 planned sites would be an appropriate number to allow for failover and guarantee adequate and convenient supply. We suggested they primarily be located in urban areas, but some should be positioned to help consumers to travel across California, safe in the knowledge that hydrogen refuelling stations were sensibly located.
We are hopeful that this comprehensive piece of analysis offered an important and a valuable contribution to the nascent hydrogen industry to increase its foothold in the California market. Modelling likely demand gives the retailer the greatest chance of success and the consumer the best opportunity to easily and readily refuel. Carrying out similar analysis and modelling will increase the likelihood of successful adoption as other states and regions look to introduce hydrogen as an alternative fuel.
Read more from the July Issue of our Fuel for Thought newsletter.