Bay of Plenty kiwifruit postharvest
Configurable line planning across several kiwifruit packhouses
Growth Mindset developed forecasting and line-planning software for a Bay of Plenty kiwifruit postharvest operator. The system used sample data and days after full bloom to estimate fruit size, then applied each packhouse's grader, equipment, labour and planning settings to prepare a site-specific production plan.
A common planning method could not assume a common packhouse
Several packhouses needed the same basic answer: what fruit was likely to arrive, what tray mix it would produce and how the line should be configured. The inputs and constraints were not uniform. Grader specifications, equipment capability, production rates, labour assumptions, cleared sizes and local priorities all changed the plan.
A useful system therefore had to separate what should be common from what had to remain local. The forecasting method could be shared. The settings that determined a feasible line plan had to stay visible and controllable at each packhouse.
Forecast fruit size before packing
The model used sample weights, pick date, full bloom date, variety and storage path to estimate fruit weight and size profile before the run. Growth curves based on days after full bloom supplied the expected distribution, which was then mapped into the commercial size bands used by the business.
Where quality-loss information was available, the system also estimated Class 1 packout. The result was an earlier view of expected trays by size and likely packout, not just a raw biological forecast.
One analytical method, with each site's operating settings kept explicit
A common calculation engine sat beneath the application. Local settings defined grader specifications, equipment capability, packing rates, labour assumptions, planning priorities and defaults.
This configuration was not a cosmetic layer. It changed which products and equipment combinations were feasible, how much manual packing was required and which options made sense for that site. Keeping these settings explicit allowed the method to remain consistent without pretending the packhouses were identical.
Convert the forecast into a site-specific line plan
The forecast profile was converted into expected trays, recommended equipment use and manual packing requirements. Optimisation logic evaluated capacity, eligibility rules, automation suitability and the priorities selected for the run.
Planners could save scenarios, review the assumptions, compare plans and export results to Excel or PDF. Structured outputs were also available for reporting and downstream automation where required.
Planning became consistent without making every site work the same way
Each packhouse began from the same forecasting method while retaining control over the settings that changed the answer. Teams could plan from an expected size profile before production began and leave a clear record of the scenario they intended to run.
The fruit size forecast also proved useful beyond line planning, extending the value of the model beyond its original role as an input to optimisation.