| Title: | Constructing Suns-Voc from Outdoor Time-Series I-V Curves |
|---|---|
| Description: | Suns-Voc (or Isc-Voc) curves can provide the current-voltage (I-V) characteristics of the diode of photovoltaic cells without the effect of series resistance. Here, Suns-Voc curves can be constructed with outdoor time-series I-V curves [1,2,3] of full-size photovoltaic (PV) modules instead of having to be measured in the lab. Time series of four different power loss modes can be calculated based on obtained Isc-Voc curves. This material is based upon work supported by the U.S. Department of Energy's Office of Energy Efficiency and Renewable Energy (EERE) under Solar Energy Technologies Office (SETO) Agreement Number DE-EE0008172. Jennifer L. Braid is supported by the U.S. Department of Energy (DOE) Office of Energy Efficiency and Renewable Energy administered by the Oak Ridge Institute for Science and Education (ORISE) for the DOE. ORISE is managed by Oak Ridge Associated Universities (ORAU) under DOE contract number DE-SC0014664. [1] Wang, M. et al, 2018. <doi:10.1109/PVSC.2018.8547772>. [2] Walters et al, 2018 <doi:10.1109/PVSC.2018.8548187>. [3] Guo, S. et al, 2016. <doi:10.1117/12.2236939>. |
| Authors: | Menghong Wang [aut]  ,
Tyler J. Burleyson [aut, cre]
, Jiqi Liu [aut]
, Alan J. Curran [aut]
, Abdulkerim Gok [aut,
ctb] , Eric J.
Schneller [aut] ,
Kristopher O. Davis [aut]
, Jennifer L. Braid
[aut] , Roger H.
French [aut, cph] |
| Maintainer: | Tyler J. Burleyson <[email protected]> |
| License: | BSD_3_clause + file LICENSE |
| Version: | 0.1.2 |
| Built: | 2024-11-05 06:35:58 UTC |
| Source: | CRAN |
This function parses the I-V curve string and creates an additional dataframe with current and voltage columns from it.
char_to_df(str)char_to_df(str)
str |
The character string to be converted to a Dataframe. |
Dataframe containing IV curve data.
char_to_df(df_wbw$ivdf[1])char_to_df(df_wbw$ivdf[1])
This dataframe contains 1 year of anonymous raw outdoor time-series IV curve data.
df_wbwdf_wbw
Dataframe with 4140 objects in 10 variables:
A local PosixCT Timestamp
IV Dataframe
Module Temperature
Plane of array (POA) irradiance
Extracted Short Circuit Current by ddiv
Extracted Open Current Voltage by ddiv
Extracted Rs (series resistance) by ddiv
Extracted Maximum Power by ddiv
Extracted Current at Maximum Power Point
Extracted Voltage at Maximum Power Point
The companion function to this one, char_to_df,
parses this string and creates an additional dataframe from it.
This function works the other way, converting that dataframe back to a character
string.
df2chr(df)df2chr(df)
df |
The dataframe, typically named "ivdf", to be converted. |
Returns a character string representing an IV curve.
df2chr(char_to_df(df_wbw$ivdf[1]))df2chr(char_to_df(df_wbw$ivdf[1]))
This function is used internally by IVxbyx and others for the calculation of 1-sun Isc values based on Isc and Irradiance measurements.
isc.1sun(isc, Irrad)isc.1sun(isc, Irrad)
isc |
Isc values |
Irrad |
Irradiance values |
Returns a calculated Isc value at 1-sun Irradiance.
isc_1sun <- isc.1sun(df_wbw$isc, df_wbw$poa)isc_1sun <- isc.1sun(df_wbw$isc, df_wbw$poa)
This function moves through IV curve data x-by-x, generating psuedo IV curves and binding them together into the output.
IVXbyX(df, corr_temp = "median", N_c)IVXbyX(df, corr_temp = "median", N_c)
df |
Dataframe containing IV data. Typically, a raw dataframe after
filtering by current accuracy based on plot of irradiance vs. temperature.
Within this package environment, use
|
corr_temp |
The temperature for all I-V features to standardize to. Pass the string "median" in order for the function to automatically calculate the median module temperature at 1 sun irradiance and use it, or input a number directly, the temperature unit should be Celsius. |
N_c |
Number of cells in series; the total number of cells in the system. |
Psuedo-IV Curve data with features extracted and evaluation parameters of fitting grouped by time periods of set length.
df <- read_df_raw(df_wbw,0.02,7) df_full <- IVXbyX(df, corr_temp = "median", 60)df <- read_df_raw(df_wbw,0.02,7) df_full <- IVXbyX(df, corr_temp = "median", 60)
This function calculates the median module temperature throughout the data to be used in corrections.
median_temp(df)median_temp(df)
df |
Dataframe containing timeseries irradiance (column name must be poa) and module temperature (column name must be modt) in unit of Celsius. |
Returns an integer value of median reported module temperature of the data.
T_corr <- median_temp(df_wbw)T_corr <- median_temp(df_wbw)
Obtain Psuedo IV Curve and other predicted IV features at reference conditions for a given week, used internally in IVxbyx function.
p_iv.week(df, temp, N_c, isc_1sun = NULL)p_iv.week(df, temp, N_c, isc_1sun = NULL)
df |
A dataframe containing timeseries I-V features of one period. |
temp |
The reference module temperature to correct the Psuedo IV curve to be, unit of the temperature should be Celsius. |
N_c |
Number of cells in series. Equal to the total number of cells in the system. |
isc_1sun |
(optional) Input an Isc 1-Sun value manually. Leave NULL to have one generated from the dataframe. |
Psuedo-IV Curve data with features extracted and evaluation parameters of fitting grouped for a single, given period.
df <- read_df_raw(df_wbw,0.02,7) df_slice <- dplyr::filter(df, df$n_period == 1) # Check that this has enough data! needs more than 10 rows to be meaningful nrow(df_slice) # needs median temperature temp <- median_temp(df_wbw) res <- p_iv.week(df_slice, temp = temp, N_c = 60)df <- read_df_raw(df_wbw,0.02,7) df_slice <- dplyr::filter(df, df$n_period == 1) # Check that this has enough data! needs more than 10 rows to be meaningful nrow(df_slice) # needs median temperature temp <- median_temp(df_wbw) res <- p_iv.week(df_slice, temp = temp, N_c = 60)
This function decouples power loss to different loss modes, specifically current mismatch, recombination, uniform current loss, and Rs loss, from the changes in IV features for a given period, used internally for power_loss_phys_bat function.
power_loss_phys(init_piv, init_isc1sun, init_prs, degr_row)power_loss_phys(init_piv, init_isc1sun, init_prs, degr_row)
init_piv |
Dataframe containing current andvolatge psuedo-IV curves of initial period, calculated from the the result of p_iv.week function applied to timeseries dataframe of initial period. |
init_isc1sun |
1 sun isc generated from the p_iv.week output from the initial period dataframe. |
init_prs |
extracted rs from the psuedo-IV curve of the initial period. |
degr_row |
one row of output from IVXbyX function. |
dataframe containing information about power loss due to various power loss modes for one given period
df <- read_df_raw(df_wbw,0.02,7) df_init <- select_init_df(df, days = 21) init <- p_iv.week(df_init, temp = 30, N_c = 60) init_piv <- data.frame(V = init$voc_corr, I = init$isc_1sun - init$isc) init_piv <- dplyr::arrange(init_piv, .data$V) init_pivf <- ddiv::IVfeature(I = init_piv$I, V = init_piv$V, crtvalb = 0.06) init_prs <- init_pivf$Rs init_isc1sun <- init$isc_1sun df_full <- IVXbyX(df, corr_temp = 30, 60) power_loss_phys(init_piv,init_isc1sun, init_prs,df_full[1,])df <- read_df_raw(df_wbw,0.02,7) df_init <- select_init_df(df, days = 21) init <- p_iv.week(df_init, temp = 30, N_c = 60) init_piv <- data.frame(V = init$voc_corr, I = init$isc_1sun - init$isc) init_piv <- dplyr::arrange(init_piv, .data$V) init_pivf <- ddiv::IVfeature(I = init_piv$I, V = init_piv$V, crtvalb = 0.06) init_prs <- init_pivf$Rs init_isc1sun <- init$isc_1sun df_full <- IVXbyX(df, corr_temp = 30, 60) power_loss_phys(init_piv,init_isc1sun, init_prs,df_full[1,])
This function decouples power loss to different loss modes, specifically current mismatch, recombination, uniform current loss, and Rs loss, from the changes in IV features.
power_loss_phys_bat(df_iv, init_df, corr_T = 40, N_c, ddiv_param = 0.06)power_loss_phys_bat(df_iv, init_df, corr_T = 40, N_c, ddiv_param = 0.06)
df_iv |
Dataframe containing psuedo-IV curves. Generate with |
init_df |
Dataframe containing first several weeks of real-world IV data.
Generate with |
corr_T |
The temperature from which to create the correction factor. Pass only numeric values. |
N_c |
Number of cells in series; The total number of cells in the system. |
ddiv_param |
Parameter passed to ddiv::IVfeature for 'crtvalb' |
dataframe containing information about power loss due to various power loss modes
df <- read_df_raw(df_wbw,0.02,7) df_init <- select_init_df(df, days = 21) df_full <- IVXbyX(df, corr_temp = 30, 60) res <- power_loss_phys_bat(df_full, df_init, corr_T = 30, N_c = 60)df <- read_df_raw(df_wbw,0.02,7) df_init <- select_init_df(df, days = 21) df_full <- IVXbyX(df, corr_temp = 30, 60) res <- power_loss_phys_bat(df_full, df_init, corr_T = 30, N_c = 60)
Given an imported datafram of Hbase-Formatted IV curve data, this function reads in the data, filters missing temperature data, and checks for a nonzero maximum power point, in case of power loss. It resets the timestamps based on the minimum timestamp, and filters Isc values for the tracer's accuracy. Finally, a n_period counter is added to the dataframe.
read_df_raw(df, tracer_accuracy, t_period)read_df_raw(df, tracer_accuracy, t_period)
df |
dataframe; the IV curve data to be filtered |
tracer_accuracy |
The accuracy of the IV tracer used. See the device's manual to find the exact value at which Isc readings are no longer accurate. |
t_period |
Data period for the Psuedo-IV curves. Addded as a column to the dataframe based on the timestamp. Use units of days, i.e. daily periods should have t_period 1, weekly periods should have t_period 7, etc. |
df_raw, a dataframe containing the raw IV curve data
df_test <- read_df_raw(df_wbw, 1, 7) # Weekly periods df_test <- read_df_raw(df_wbw, 1, 1) # Daily periodsdf_test <- read_df_raw(df_wbw, 1, 7) # Weekly periods df_test <- read_df_raw(df_wbw, 1, 1) # Daily periods
The power_loss_bat function makes use of a dataframe containing the first several days of psuedo-IV curves. This function creates that dataframe for the user from the raw dataframe containing IV Curve measurement data.
select_init_df(df_raw, days = 21)select_init_df(df_raw, days = 21)
df_raw |
The raw dataframe containing request input columns the same as the example df_wbw dataset and the column of day, which starts from 1 and column of n_period, which is the index of the period, decided by how many days to be grouped as one period, the column of day and n_period can be generated from the function read_df_raw |
days |
The number of initial days to subsample. Default: 21 |
Subset of input dataframe within the first several days decided by the input parameter "days".
df <- read_df_raw(df_wbw, 1, 7) df_init <- select_init_df(df, days = 21)df <- read_df_raw(df_wbw, 1, 7) df_init <- select_init_df(df, days = 21)
This method uses a physical model to correct Voc to certain reference conditions. Since indoor Suns-Voc is conducted at a steady 25C, a correction on the outdoor readings is necessary for meaningful comparison.
voc.corr(df)voc.corr(df)
df |
A dataframe time series data with columns of voc, lnSun and modt, the dataframe is converted from the request input dataframe like df_wbw by the piv_iv_week function before using voc_corr function. |
returns a list object of Voc model.
df <- read_df_raw(df_wbw,0.02,7) # subset data to first period df_slice <- dplyr::filter(df, df$n_period == 1) N_c <- 60 # true of the example data. N_c is the number of cells in series df_slice <- dplyr::mutate(df_slice, T_K = .data$modt + 273.15, lnSun = 1.38e-23/1.6e-19 * N_c * .data$T_K * log(.data$isc), isc2 = .data$isc^2, T_lnIsc2 = .data$T_K * .data$lnSun * log(.data$isc), rs = as.numeric(.data$rs), expVoc = exp(-.data$voc/.data$T_K) / .data$isc, I0 = (N_c * 1.38e-23 * .data$T_K) / (1.6e-19 * .data$isc)) voc_mod <- voc.corr(df_slice)df <- read_df_raw(df_wbw,0.02,7) # subset data to first period df_slice <- dplyr::filter(df, df$n_period == 1) N_c <- 60 # true of the example data. N_c is the number of cells in series df_slice <- dplyr::mutate(df_slice, T_K = .data$modt + 273.15, lnSun = 1.38e-23/1.6e-19 * N_c * .data$T_K * log(.data$isc), isc2 = .data$isc^2, T_lnIsc2 = .data$T_K * .data$lnSun * log(.data$isc), rs = as.numeric(.data$rs), expVoc = exp(-.data$voc/.data$T_K) / .data$isc, I0 = (N_c * 1.38e-23 * .data$T_K) / (1.6e-19 * .data$isc)) voc_mod <- voc.corr(df_slice)