Here are two ways to recreate your plot using dataset1.

1. Scale in proportion to Twh. This seems simplest and most efficient, provided you don’t need the visible bars to be composed of many stacked smaller bars.

ggplot(dataset1, aes(
  x = Tot*Twh, 
  y = reorder(Country, Tot), 
  fill = Energy_Source
)) +
  geom_col()

2. tidyr::uncount is what you want if you want to make copies of each observation. This replicates your dataset2 approach. I have added borders to show how this makes many small bars that are stacked together. This approach is fine here, but I’ve had issues where it might plot very slowly (e.g. if >100k observations to plot), or plot messily (e.g. if the borders overwhelm the areas or create moire effects), or inefficiently (e.g. a vector format like PDF would save a separate object for each bar plotted, even if <1 pixel).

ggplot(dataset1 |> tidyr::uncount(Twh), aes(
  x = Tot, 
  y = reorder(Country, Tot), 
  fill = Energy_Source
)) +
  geom_col(color = "gray50")

Edit: For the OP’s description of “sequential” data, I think it’s more efficient computationally and cleaner to plot if you calculate the aggregations (here, the average usage across years of energy source per country) with a dplyr step.

Compare a version of what’s in the OP’s suggested answer:

rbind(dataset1, dataset1b) |>
  mutate(across(Twh, ~.x * 10)) |>
  uncount(Twh) |>
  ggplot(aes(
    x = Tot / (10 * 2), 
    y = reorder(Country, Tot), 
    fill = Energy_Source
  )) +
  geom_col()

…to a version using dplyr. Same superficial visual appearance, except the uncount version plots 4,400 observations, vs. the dplyr version just plots the 9 contiguous bars we can see.

rbind(dataset1, dataset1b) |>
  summarize(Total = mean(Tot * Twh), .by = c(Country, Energy_Source)) |> 
  ggplot(aes(
    x = Total,
    y = reorder(Country, Total), 
    fill = Energy_Source
  )) +
  geom_col()

For reference, your example plot with the same dimensions:

Jon Spring answer is great.
/users/6851825/jon-spring?tab=profile

Now, to go further if you have a sequential dataset with decimal values to the variables you are working with (and you do not want to round them), here is how to deal with it:

Example dataset:

dataset1 <- data.frame(
  "Country" = c(rep(x = "Italy", times = 4), rep(x = "France", times = 5)),
  "Energy_Source" = c("Biofuel", "Nuclear", "Gas", "Hydro", "Biofuel", "Nuclear", "Gas", "Hydro", "Wind"),
  "Twh" = c(24.3, 15.7, 40.0, 20.0, 19.1, 75.9, 10.0, 5.0, 10.0),
  "Tot" = c(rep(x = 100, times = 4), rep(x = 120, times = 5)),
  "Year" = c(rep(x = 2012, times = 9))
)

dataset1b <- data.frame(
  "Country" = c(rep(x = "Italy", times = 4), rep(x = "France", times = 5)),
  "Energy_Source" = c("Biofuel", "Nuclear", "Gas", "Hydro", "Biofuel", "Nuclear", "Gas", "Hydro", "Wind"),
  "Twh" = c(25.0, 15.0, 40.0, 20.8, 19.2, 75.0, 10.0, 5.0, 10.0),
  "Tot" = c(rep(x = 100, times = 4), rep(x = 120, times = 5)),
  "Year" = c(rep(x = 2013, times = 9))
)

dataset1 <- rbind(dataset1, dataset1b)
head(dataset1, n = 10)

##    Country Energy_Source  Twh Tot Year
## 1    Italy       Biofuel 24.3 100 2012
## 2    Italy       Nuclear 15.7 100 2012
## 3    Italy           Gas 40.0 100 2012
## 4    Italy         Hydro 20.0 100 2012
## 5   France       Biofuel 19.1 120 2012
## 6   France       Nuclear 75.9 120 2012
## 7   France           Gas 10.0 120 2012
## 8   France         Hydro  5.0 120 2012
## 9   France          Wind 10.0 120 2012
## 10   Italy       Biofuel 25.0 100 2013

Note: Tot represents the sum(Twh) by Country

First, multiply the Twh column with a round number that will transform all of the decimals to int. In the example, none of the values has more than one number after the 0, so we multiply Twh by 10, let’s called it multiplicator

multiplicator <- 10
dataset1$Twh <- dataset1$Twh * multiplicator
head(dataset1)
##   Country Energy_Source Twh Tot Year
## 1   Italy       Biofuel 243 100 2012
## 2   Italy       Nuclear 157 100 2012
## 3   Italy           Gas 400 100 2012
## 4   Italy         Hydro 200 100 2012
## 5  France       Biofuel 191 120 2012
## 6  France       Nuclear 759 120 2012

Second, we apply the tidyr::uncount() function to dataset1 with Twh as a parameter, it is the variable that will dupplicate each row according to its value at each row.

dataset1 <- dataset1 |> tidyr::uncount(Twh)
##   Country Energy_Source Tot Year
## 1   Italy       Biofuel 100 2012
## 2   Italy       Biofuel 100 2012
## 3   Italy       Biofuel 100 2012
## 4   Italy       Biofuel 100 2012
## 5   Italy       Biofuel 100 2012
## 6   Italy       Biofuel 100 2012

Finally we plot the data using ggplot2 as following:

ggplot(dataset1, aes(
  x = Tot / (multiplicator * length(unique(dataset1$Year))), 
  y = reorder(Country, Tot), 
  fill = Energy_Source
)) +
  geom_col()

See:

As ggplot2 understands the proportion of each element by their occurence in a dataset and that multiplicator and the non grouping column wich is Year affected the occurence, we divide Tot by (multiplicator * length(unique(dataset1$Year))).