Using the software program ries
I easily found that
$$e^\pi-\pi \approx 19.999099979 \approx 20$$
which I think is more accurate and interesting than
$$ e^3 \approx 20.085536923 \approx 20.$$
The approximation with $\pi$ is already known and is
asked about in MSE question 724872 (Thanks
to a comment by user 'Jam').
Note: Here is output from ries 2.99573227355399099
:
Your target value: T = 2.99573227355399 mrob.com/ries
x = 3 for x = T + 0.00426773 {30}
x-pi = 1/-7 for x = T + 0.00300324 {66}
sqrt(ln(x)) = pi/3 for x = T - 0.00169507 {71}
e x = 5+pi for x = T - 0.000607718 {70}
log_5(x) = 1-1/pi for x = T - 0.000152139 {77}
e^x = 4*5 ('exact' match) {65}
(3"/x)^2 = 1/ln(phi) for x = T - 5.12827e-05 {84}
e^x+pi = e^pi for x = T - 4.50021e-05 {73}
1/(pi-x) = phi^4 for x = T - 3.76537e-05 {87}
sqrt(e^x) = 2 sqrt(5) for x = T + 4.44089e-16 {80}
(Stopping now because best match is within 2.66e-15 of target value.)
Note: You may be interested in the following
$$ e^{\pi\sqrt{22}} \approx 2508951.99 $$
$$ e^{\pi\sqrt{37}} \approx 199148674.9999 $$
$$ e^{\pi\sqrt{58}} \approx 24591257751.999999822 $$
which have a satisfying explanation.
The biggest of these seems to be
$$ e^{\pi\sqrt{163}} \approx
262537412640768743.99999999999925 $$
For details read the Wikipedia Heegner number
article.