# Notes PV Home Charging paper

## rebuttle

## open questions
- [ ] Think about excluding 0656f4967b02fa467e7a85ea85b62cd1 for having only 70 kwh of total demand
- [x] Two users have very large houses (2*37 + 45 kwp). Should we exclude them or limit the charging schedule? (They are excluded in the current version)
- [x] How should we describe the number of users?
% Full user journey of SBB GC users: SBB GC has theoretically 143 users (some of which never made any recording)
% 143 users
% 126 can be assigned to a house available in the GWR
105 are good matches
of which 71 live in single familiy homes
11 live in two-familiy homes
4 live in multi-familiy homes with known nb of hh
85 are good matches in well recognized homes but 3 can't be assigned to cars (probably they dropt out of GC)
-- > 82 users (-3 that are almost never at home)
-> Only describe a summary of the criteria and the final number of users
- [x] Nice scenario names
- [x] matching_bmw_to_address.csv had some doublons. Some people gave wrong addresses and we (maybe Rene back in the day) corrected them. I updated the matching_bmw_to_address.csv with the coordinates of the new houses.
- [ ] PV naming convention:
- Good:
- PV energy generation
- Solar energy generation
- Solar power generation
- PV/solar electricity generation
- Bad:
- PV generation
- Solar generation

## Open Todo's
- [x] For raw SOC plot (figure 2), the legend should be outside of the plot
- [x] exclude last two weeks of December (I excluded 8 days), the last day is not the 23rd
- [x] Flowchart: Maybe add scenario name into headers
- [x] Figure 9: Names in legend do not start with a capital letter
- [x] Figure 10: Y-axis has no unit
- [x] Add SoC_ref to scenario description (scenario 1)
- [x] PV Panel efficiency: This is the overall system efficiency. output (system) /Input (sonne)
- [x] BMW Mobiltiy behavior plot: Disable interpolation
- [x] Add numbers for swiss transport sector
- [ ] [HM] Uncertainty intervals are for the mean/median not for the data (only something to check in the description of the plots)
- [ ] [HM] Redo Sensitivity for roof size and total consumption (meaning the description)
- [x] [HM] Repair citations (author etc.)
- [ ] [HM] Mention garages/no garages in the matching
- [ ] [HM] Renes plotting function stores some data. Find out what data this is exactly (because of privacy) and delete it if necessary.
- [ ] [RB] Describe panel properties
- [ ] [RB] PV system (overview kind of a flow chart) (Do we need to describe that or can we just cite your paper?)
## Things to keep in mind for journal
- https://www.journals.elsevier.com/renewable-and-sustainable-energy-reviews
- [ ] Add SDGs in the introduction "Sustainability - the United Nations Sustainability Development Goals (SDGs)"
- Word limit is 4000 - 8000 words. We currently have 7200 (rather at the upper bound)
- We should not use line numbers

## tasks for experiment
- [x] Implement factor to correct roof area (if we include 2-familiy homes)
- [ ] !Needed by car can be greater than 100 SOC (This is intended because 1 segment where the user is not at home could have several charging cycles). The calculation of the coverate should not consider this.
- [x] Sensitivity analysis with panel type (efficiency)
- [x] Plot for when EVs are at home. Matrix heat-map plot, hour of day X time of year (color is % at home)

## tasks for analysis of results

other

## thinks to keep in mind / to mention in the paper
- [x] Total battery capacity of the vehicles
- [x] Rename scenarios and give them descriptive names (e.g., Baseline = real charging, scenario 1: in-segment optimization, ...)
- [x] Describe the battery storage (What should this mean?)
- [x] Switching from EV (electric vehicle) to BEV (battery electric vehicle, as opposed to [PHEV](https://www.pkw-label.de/alternative-antriebe/elektrofahrzeuge-bevphevreev)). (HM: I think we need at least think about what we are implying by using EV instead of BEV)
- [x] What to write as CO2? Probably 49.91 g CO2-eq/kWh would be the most correct?
- [x] Roof validation: Mention Garage/non-garage

## postponed for camera ready:
- [ ] Fixed color scheme for scenarios (postponed for camera ready)

## prio 2 / postponed
- [ ] Co2 Emission reduction per user (scatter plot, new vs. old)
- [ ] Quantify PV overgeneration: This could be a problem for the grid
- [ ] Include demographics from SBBGC final report. (probably not)
- [ ] Add a section that tells how to reproduce results in the readme (could include a workflow). Maybe we could have some dummy mobility data?
- [ ] Sensitivity to minimum charge (e.g., if below 50 % charge from grid) (a lot of work, we keep this for the review)
_____________________

## done
- [x] CO2 plot: For the storage capacity we should include the emissions of the home storage production. Maybe there is an optimal size!
- [x] Find correct values for swiss co2 mix + co2/kwh for pv rooftop

[TOC]

#### old todos /done (already scanned for useful notes)
- PV data exploration ( We want to give the reader an impression of the solar data that we are using. THerefore we have two possibilities)
- Possibility 1: We show aggregated (maybe + scatter) power generation over the year (just as the pv plot)
- Possibility 2: We focus on the resolution of the used pv model. Main arguments against the first possibiltiy are that it could give the impression that we are using low resolution data and that the variation of pv over the year is very well known.
-

- [x] Experiment effect on CO2 emissions:
- [x] For every scenario: Get the amount of energy that was home charged vs. the amount of energy that was charged externally.
- [x] Multiply home charging with roof top pv emission factor and externally charged with swiss power system factor.
- For the storage capacity we should include the emissions of the home storage production. Maybe there is an optimal size!
- Plots
- [x] Charging efficiency (SoC -> kwh)
- [x] Here the correct formula is missing
- [x] At the moment we fit the regression without intercept following a physics argument (when dSoC is zero, then dkwh has to be zero)
- [x] PV efficiency is not used? (our model is a PV panel that already has this included)

- Code
- [x] push data preprocessing code [HM]
- [x] [JH] Limit the max charging power to 11 kw
- [x] [JH] Create consistent solar model structure. There are two identical folders with data&code in `data/solarrad` and `src/solarrad`. The Folders are probably the same as jannik.pvmodel2.pv_swissbuildings_json
- [x] [JH] check which version of the folders is the _correct_ version
- [x] [JH] seperate data and code and sort them into `./src` and `./data`
- [x] [JH] Change battery specifications to [tesla powerwall](https://www.tesla.com/de_ch/powerwall)
- [x] [JH] integrate `jannik` folder into `src`
- writing
- [x] [HM] update on local consumption for EV literature research
- [x] [HM] Section 2 Background
- Data
- [x] [HM] We need a (synced) place where we can exchange data to ensure that we are all using the same state
- [x] We need a (synced) place where we can exchange data to ensure that we are all using the same state [HM]

- [x] [RB] What PV model did we use? (its the solar panel thing)

# Meeting Notes

## 29/09/2020
#### Open questions for meeting:
- what data can we publish
- can be published
- json of solar radiance
- can not be published
- anything with user id
- BMW data

## 00/02/2020

Omit everything that is related to motion tag tracking data
### PV scenarios
- We consider several pv generation scenarios in our paper. All of them consider a different conversion function between solar radiation to generated power. We have 3 non-linear scenarios that are considering specific pv panel models and a non-linear converter. Additionally we do a sensitivity analysis for different constant power factors (27 scenarios).
- We choose 1 main scenario (the medium pv panel) and 1 sensitivity analysis for the paper. The rest is in the attachment.
- Move _EV Energy Demands_ and _Photovoltaic Potential_ into the data section as it is a presentaion of the data set ?

### open todos prio 1

- [ ] ~~Optimize i/o of new pv data (RB)~~
- [ ] ~~Check single vs. multi family home classification
We assign a number to every building with the estimated number of households (simplified e.g. 1=1 household, 2=2 households, 3=more, 4=garbage)~~
- [ ] ~~Rene prepares the data (RB)~~
- [ ] ~~Jannik does the classifaction (JH)~~
- [ ] ~~Limit the max charging power to 11 kw (JH)~~

## 23/01/2020
### Notes
- Query for bmw data with home flag: (Database: sbb-green on go-eco server)
select * from ev_homepv.ecarid_is_athome limit 10

### Todo's
- Henry finishes code + data export
- Jannik works on new scenarios (Baseline + scenario 2)
- Work on paper (DB, HM) ;-)

## 21/01/2020
### Notes
- For now we assume 11 kw home chargers

### Todo's
- Find out about max charging power of BMWs
- Think if max charging power is a problem
- Implement scenario 2 (JH)
- Export new data with consumption per trip (HM)
- For baseline: All load entries that are at home
- For Scenario 2: consumption per trip
- Work on paper (DB, HM) ;-)

# Technical stuff

## Scenario design
### Baseline 1: What the user really did
We use the true charging curve from the data and compare it to the available pv power

### Baseline 2: Home --> plug-in (optional)
When the user comes home, he plugs in and charges (limited by max charging power)

### Scenario 1: Optimal 1-block charging
_Within 1 block_, compare the energy that was used for charging with the generated pv energy.

### Scenario 2: Optimal multi-block charging
A user charges as much pv energy as possible and at the end of each segment we compare the current state of charge with the energy that is required for the next trip. Missing energy is charged with non-pv power (at anytime in the segment).
For each trip, we know the total required energy of the trip (might be > 100 % SOC because he might have charged the car externaly) and we substract the total required energy from the current SOC. If the SOC is < 0 %, then the rest was charged externally using non-pv power.

### Scenario 3: Scenario 2 + Battery
In parallel, we simulate the SOC of a home-storage. The car get's charged by pv first and then by battery (which only got charged by pv)

# Figure guidelines:
- 1 independent function per plot
- We use Rene's plotting function
- No title
- Matplotlib + wrappers (seaborn, pandas, etc)

# Code guidelines
- All code is in the `./src`
- The folder `./src` is the working directory when executing code
- If possible use plattform independent and relative paths (e.g., `os.path.join('data', 'solarrad')`)
- All data is in `./data`
- All data that is user specific (=anything that gives non-aggregated information of individual EVs) musst enter the git
- PV data can be in the repo (check with Rene before publication)
- Figures can be in the repo

# assumptions and limitations
- Battery is weather independent
-