Yokohama: MATSim Application for Resilient Urban Design

In Yamagata and Seya (2015), we proposed the concept of a resilient local electricity-sharing system as a complement, or alternative, to a FIT (feed-in tari ) to achieve CO2-neutral transportation in cities. In our proposed system, electricity generated from widely introduced solar PVs (Photovoltaic Panels) is stored in cars “not in use” in a city. In Japan, almost half the central Tokyo metropolitan area cars are used only on weekends and thus are kept parked weekdays. These cars could represent a huge new storage potential if they were replaced by EVs; that is, they could be used as storage batteries in a V2C (Vehicle to Community) system. This study analyzed the potential of EVs as storage batteries in emergency cases. Speci cally, we focused on the following three questions:


Introduction
In Yamagata and Seya (2015), we proposed the concept of a resilient local electricity-sharing system as a complement, or alternative, to a FIT (feed-in tari ) to achieve CO 2 -neutral transportation in cities. In our proposed system, electricity generated from widely introduced solar PVs (Photovoltaic Panels) is stored in cars "not in use" in a city. In Japan, almost half the central Tokyo metropolitan area cars are used only on weekends and thus are kept parked weekdays. These cars could represent a huge new storage potential if they were replaced by EVs; that is, they could be used as storage batteries in a V2C (Vehicle to Community) system. This study analyzed the potential of EVs as storage batteries in emergency cases. Speci cally, we focused on the following three questions: 1. How much residential demand can be met (in each 24 hour) by electricity from just PVs, which are installed on the roofs of all detached houses in the study area?
2. How many EVs are needed to store all surplus electricity (PV supply minus demand)?
3. How does EVs driving change the load curve and how can mass-adopted PVs ful ll total demand?
To answer our second and third questions, we needed to know (a) the number of cars parked at home during each hour (that is, the time each car arrived at home a er use) and (b) the amount of battery charge consumed by each driver during his/her daily trips (that is, trip duration). For this simulation, we used MATSim. In this chapter, we brie y introduce our MATSim application for a local electricity-sharing system in Yokohama city, based on Yamagata and Seya (2013); Yamagata et al. (2014,2015).

Results
We assumed that PV was installed on the roof of each detached house in Yokohama city. Then, we calculated the amount of electricity supplied each hour throughout the whole day by employing simple intensity method. The O-D trip data used are from the Fourth Person Trip Survey in Tokyo Metropolitan Area, implemented in 1998. The data are available through the People Flow Project (http://pflow.csis.u-tokyo.ac.jp) on request (application) and include the O-D trips by tra c mode, time of day, purpose, etc. for each micro district, called cho-cho-moku. The Person Trip survey is a national survey that focuses on people's travel behavior during a given few days of each month, from October to December. Because the number of cars in Yokohama for each chocho-moku was unknown, the city-level value was allocated to the cho-cho-moku (areal weighting) and adjusted for the size of the population. The road-network information was taken from the National Digital Road Map Database and included su cient data on road capacity, width classication, link length, number of lanes and travel speed to perform tra c simulations in MATSim. MATSim requires a daily "plan le" for each agent (car driver); we prepared these les by using the Fourth Person Trip Survey, which captured the daily movements of 722 000 people. Because the Fourth Person Trip Survey sampled approximately 2 % of the population of the Tokyo metropolitan area, the plan le was replicated according to the intensity factor provided by the People Flow Project, resulting in 505 335 agents. From the MATSim simulation, we had obtained each agent's trip duration and arrival time.
Considering load curve changes due to the EVs driving, we then asked if massively adopted PVs would be enough to satisfy total energy demand in Yokohama. In Figure 96.1, the solid and dashed lines represent electricity surplus cumulative distribution, charged to or discharged from the batteries of EVs, not in use and used only for charging the EVs in use, during May and August (solid line, maximum; dashed line, average). The dotted line in the gure represents the scenario where electricity surplus was both charging EVs and satisfying households' typical electricity demand  Figure 6) under maximal/average solar irradiance. However, in August (high demand, high PV supply), the electricity surplus was su cient for charging EVs, but not enough to meet the households' huge electricity demand due to evening use of air-conditioning.
To meet household electricity demand, PV electricity needs be e ciently stored in EVs and locally shared. For example, if a high-a ordability zone (storage capacity is greater than electricity surplus) is adjacent to a low-a ordability zone (storage capacity is smaller than electricity surplus), then the share of their EV capacity increases the ratio of stored PV electricity. Because storage a ordability (storage capacity minus electricity surplus) is signi cantly di erent regionally (see Figure 96.2), clustering of community-based local sharing must be carefully designed. In this study, we attempted to optimize community clusters using several di erent algorithms. Firstly, the number of clusters was assumed 18 to be the same as the number of Yokohama city wards. Then, cluster optimization was performed by minimizing (the sum of storage a ordability in the 18 clusters) plus k (minimum circularity in these clusters), where k was the weight for the circularity. The rst term balanced storage capacity and electricity surplus to increases the rate of stored PV electricity; the second term decreased inter-point distance within each cluster, as well as electricity sharing (transmission) cost. The minimization was conducted in every month through a simulated annealing algorithm to nd optimal spatially clustered communities. Figure 96.3 shows four-month clustering results; all clusters indicate positive storage a ordability in April, May, June, July, September, and October. In other words, PV electricity covers whole household electricity demands, if EV capacities are shared with these optimized clusters. In summary, we applied MATSim to analyze the potential of EVs in a V2C system and found that EVs can cover typical household electricity demands in some months and the cover ratio can be increased by community clustering for local electricity sharing. In the future study, we plan to use MATSim to simulate mobility behavior for electricity sharing community scenarios and extend our clustering analysis utilizing simulated behavior. Finally, development of community level mobility sharing service would be a very important topic to integrate MATSim simulations with our land use and transportation scenarios, such as compact and dispersion scenarios (see Yamagata et al., 2013).