while loop and double acting weirdly

Question

Trying to solve a differential equation numerically (for the first time). My program asks for a bunch of parameters and then calculates the amount of time required to chill a bottle of wine to a given temperature. This is my main:

import java.util.Scanner;

public class Step2_lab11 {
    public static int TAO = 50;
    public static double DELTA_MINUTES = 0.1;

    public static void main(String[] args) {

        System.out.println("VINETS AVKYLNINGSTID \n");
        System.out.println("Ange vinets temperatur:");

        Scanner userIn = new Scanner(System.in);
        double wineTemp = userIn.nextDouble();

        System.out.println("Vinets önskade temperatur:");
        double preferredTemp = userIn.nextDouble();

        System.out.println("Kylens/frysens temperatur:");
        double chillTemp = userIn.nextDouble();

        WineChiller wineChiller = new WineChiller();

        double elapsedTime = 0.0;

        while(wineTemp > preferredTemp) {

            elapsedTime = elapsedTime + DELTA_MINUTES;
            double dT = wineChiller.getChillingTime(TAO, DELTA_MINUTES, chillTemp, preferredTemp, wineTemp);
            wineTemp = wineTemp - dT;
            System.out.println(elapsedTime);


        }
    }

}

And this is the WineChiller.java file:

public class WineChiller {
    public WineChiller() {


    }

    public double getChillingTime(int TAO, double DELTA_MINUTES, double chillTemp, double preferredTemp, double wineTemp) {

        double dT = (wineTemp - chillTemp) * DELTA_MINUTES  / TAO;
        return dT;
    }

}

The Syso part of the while loop produces this (with wineTemp = 25, preferredTemp = 16, chillTemp = 5)

0.1
0.2
0.30000000000000004
....
29.300000000000146
29.400000000000148
29.50000000000015
29.60000000000015
29.700000000000152
29.800000000000153
29.900000000000155

No idea why it adds the random decimals. I also think (but I'm not 100%) the right answer should be exactly 30 minutes, not 29.9. Am I missing some obvious logic error here?

Answer 1

You need to read this.

This is just how binary numbers and IEEE floating point representation work.

You can no more represent 0.1 exactly in binary than you can 1/3 in decimal.

This is why you should not compare values when you use double or float; you need a tolerance for absolute value of differences.

Looks like a simple Euler integration of the first order ODE for transient heat transfer for a lumped mass. Make sure that you understand how time step choice affects stability and accuracy.