How to update ARIMA forecast in R? I have a time series data of 30 years and found that ARIMA(0,1,1) has best model among others. I have used the simulate.Arima (forecast package) function to simulate the series into the future.
library(forecast)

series <- ts(seq(25,55), start=c(1976,1))

arima_s <- Arima(series, c(0,1,1))

simulate(arima_s, nsim=50, future=TRUE)

Later on, i have found the updated value of first forecasted year (i.e. series[31] <- 65). Now i want to simulate the series with this updated value. I am wondering how to do this in R.
 A: Update: 
It turns out that the Arima function has an argument for supplying old model:
adj_s <- Arima(series,model=arima_s)

The end result might be the same for both approaches, but I would advise using second one, because it clearly is tested more thoroughly.
**Old answer: **
As it happens, I encountered similar problem recently. Here is the function which takes existing arima model and applies it to new data. 
adjust.amodel <- function(object,extended) {
   km.mod <- makeARIMA(object$model$phi,object$model$theta,object$model$Delta)
   km.res <- KalmanRun(extended,km.mod)

   object$x <- extended
       object$residuals <- ts(km.res$resid,start=start(extended),end=end(extended),frequency=frequency(extended))
       object$model <- km.mod
   object
}

In your case here is how you should use it:
series[31] <- 65
adj_arima_s <- adjust.amodel(arima_s,series)
simulate(adj_arima_s, nsim=50, future=TRUE)

The usual caveats apply though. You need to have good reasons to do that. If the data changes this means that the model should change, so what you are doing is ignoring the new information and sticking to the old model, which might be the wrong one. You can compare this to producing the model fit by plucking the model coefficients out of the blue air. Although the coefficients have statistical validation, it comes from the old data, so interpretation of the results should take this into consideration.
