R/refineryLP.R

In RTL: Risk Tool Library - Trading, Risk, 'Analytics' for Commodities

#' LP model for refinery optimization
#' @description Plain vanilla refinery optimization LP model.
#' @param crudes Data frame of crude inputs. `tibble`
#' @param products Data frame of product outputs and max outputs. `tibble`
#' @returns Optimal crude slate and profits. `tibble`
#' @export refineryLP
#' @author Philippe Cote
#' @examples
#' refineryLP(crudes = RTL::refineryLPdata$inputs, products = RTL::refineryLPdata$outputs)
refineryLP <- function(crudes = RTL::refineryLPdata$inputs, products = RTL::refineryLPdata$outputs) {
  if (!requireNamespace("lpSolve", quietly = TRUE)) {
    stop("Package \"lpSolve\" needed for this function to work. Please install it.",
      call. = FALSE
    )
  }

  GPW <- tibble::tibble(
    Element = c("Gross.Product.Worth", "Crude.Cost", "Processing"),
    LightSweet = c(
      products %>% dplyr::transmute(LightSweet = prices * LightSweet.yield) %>% stats::na.omit() %>% sum(),
      crudes %>% dplyr::filter(info == "price") %>% dplyr::select(LightSweet) %>% as.numeric(),
      crudes %>% dplyr::filter(info == "processing.fee") %>% dplyr::select(LightSweet) %>% as.numeric()
    ),
    HeavySour = c(
      products %>% dplyr::transmute(HeavySour = prices * HeavySour.yield) %>% stats::na.omit() %>% sum(),
      crudes %>% dplyr::filter(info == "price") %>% dplyr::select(HeavySour) %>% as.numeric(),
      crudes %>% dplyr::filter(info == "processing.fee") %>% dplyr::select(HeavySour) %>% as.numeric()
    )
  )
  GPW <- GPW %>%
    dplyr::select(-Element) %>%
    colSums()
  constraints <- products %>% dplyr::select(product, LightSweet.yield, HeavySour.yield, max.prod)
  out <- lpSolve::lp(
    direction = "max",
    objective.in = as.numeric(GPW),
    const.mat = as.matrix(constraints[, 2:3]),
    const.dir = c(rep("<=", 4)),
    const.rhs = as.matrix(constraints[, 4]),
    all.int = FALSE, compute.sens = 1
  )
  out <- list(profit = out$objval, slate = out$solution)
  return(out)
}