A limited mock election showed Minneapolis election officials that ranked-choice voting is doable this fall, but they warned that some race results could be significantly delayed because more hand-counting is usually required.

Election officials estimate they'll need 10 to 20 days to determine winners in all races, Election Director Cindy Reichert told the Minneapolis City Council on Thursday. And that's with counting around the clock.

Results should be apparent on election night in contests where voters' machine-counted first preferences are enough to win. That's most likely in races in which an incumbent or party-endorsed candidate faces token opposition.

But in the most evenly contested races, especially those where voters fill several seats on a board, the results may not be known until close to Thanksgiving because first preferences may not determine a winner. In that case, voters' second or even third preferences would need to be tallied by hand.

Reichert said they'll release unofficial totals election night that show where the candidates rank on first preferences. But experience elsewhere has shown that candidates trailing after first-preference votes sometimes leapfrog to the top on second-preference votes.

Ranked-choice voting, also known as instant-runoff voting, permits voters to rank their top three candidates for a seat in order of preference, bypassing a primary. Candidates must hit a threshold to be elected. That's over 50 percent of ballots for a single seat such as mayor or council member, but 25 percent for citywide Park Board seats and 33 percent for Board of Estimate and Taxation.

Candidates can hit that threshold with first preferences, which machines can count, or with the second or third preferences of voters who chose an eliminated candidate as their first choice.

Reichert said that hand counting wouldn't likely begin until two days after the election, because of procedures needed to prepare the ballots.

The citywide Park Board race is considered among the most likely to need a hand tally. That's because it already features three incumbents plus a former commissioner who, with less well-known candidates, could splinter the vote such that nobody hits the required threshold in the machine count.

The council decided to use ranked-choice voting in November, despite having machines capable of counting only first preferences. So if first preferences don't produce a winner, the city will use election judges and possibly other city workers for a hand count, following procedures still being refined.

The city made backup plans in case the Minnesota Supreme Court reverses a trial court that upheld the law. The higher court is expected to rule by June 11. That's the city's deadline for deciding whether to revert to traditional voting.

Space will limit the city's ability to count, Reichert told the council. The test found that only three council or park district races could be counted at once in the election warehouse, unless more space is leased there. Since all races are on one ballot, citywide races can't be counted simultaneously, Reichert said.

In general, the more candidates file for each city office during the July 7-21 filing period, the fewer votes any one candidate is likely to get.

In San Francisco, four of seven single-seat local contests last fall required counts that went beyond voters' first choices. That also was true for three of eight single-seat races in Pierce County, Washington.

Steve Brandt • 612-673-4438