Skip to content

FOOD x hyperion

Griffith Foods: Smart Algorithms to Boost Efficiency

Grif­fith Foods, a lead­ing US com­pa­ny spe­cial­is­ing in food ingre­di­ents for major brands like McDon­ald’s, is expe­ri­enc­ing a sig­nif­i­cant surge in demand for its prod­ucts. To meet this grow­ing demand, they are con­stant­ly seek­ing inno­v­a­tive solu­tions to boost their pro­duc­tion capac­i­ty. In response, Hype­r­i­on devel­oped a smart algo­rithm that enables Grif­fith Foods to reclaim 17 pro­duc­tion days, opti­mis­ing their effi­cien­cy and help­ing them scale up to meet the increas­ing mar­ket needs.

  • Food Industry
  • Production Efficiency
  • Smart Algorithm

A collaboration between

  • Arinti
  • Numble
Manufacture worker packing tomatoes conveyor machine with vegetables closeup Two team members at Griffith Foods working together on a laptop, discussing data on the screen in a modern office setting.

Challenge

As phys­i­cal expan­sion of facil­i­ties was not an option, Grif­fith Foods had to look for oth­er ways to pro­duce more. A pain point was the pro­duc­tion line down­time dur­ing clean­ings. To assure the qual­i­ty of each prod­uct and avoid cross-con­t­a­m­i­na­tion, such as mix­ing dyes or aller­gens, the pro­duc­tion line must be com­plete­ly cleaned between var­i­ous prod­uct types.

This process is known as wet clean­ing. It takes 30 to 90 min­utes, often bring­ing pro­duc­tion lines to a stand­still. Down­time means lost sales, so the need was high to reduce this clean­ing time. On the pos­i­tive side, exten­sive wet clean­ing is not always nec­es­sary. If prod­uct types are com­pat­i­ble, a quick ​‘dry clean­ing’ is enough, where the machines are hosed down with high air pressure.

Services used

AI Expertise

Data Science & Machine Learning

Solution

To reduce clean­ing time, our spe­cial­ists wrote an arti­fi­cial intel­li­gence algo­rithm based on Grif­fith Foods’ prod­uct data. The algo­rithm uses this data to deter­mine an opti­mal prod­uct sequence based on prod­uct sim­i­lar­i­ties and dif­fer­ences and forms sequences of prod­ucts that only need dry clean­ing. This great­ly reduces the num­ber of longer wet clean­ings, which only need to be done between incom­pat­i­ble sequences.

Implementation

To make sure that Grif­fith Foods did not invest in an arti­fi­cial intel­li­gence mod­el that would not deliv­er the best results, we first man­u­al­ly col­lect­ed and analysed data on key fea­tures that influ­ence the plan­ning process, assist­ed by Grif­fith Foods’ own plan­ning experts. After iden­ti­fy­ing these crit­i­cal fac­tors, our AI experts select­ed poten­tial mod­els and did sev­er­al rounds of test­ing and fine-tun­ing. This man­u­al method helped us nar­row down to two mod­els that per­formed well. These mod­els could be fur­ther trained for even bet­ter results and in the end, we chose the best-per­form­ing mod­el for implementation.

The oper­a­tors at Grif­fith Foods played a cru­cial role in this imple­men­ta­tion process by shar­ing their spe­cif­ic knowl­edge. They have years of expe­ri­ence in flag­ging the para­me­ters that affect plan­ning, so their exper­tise was invalu­able in refin­ing the mod­els. Addi­tion­al­ly, the eval­u­a­tion of ​“which mod­el per­forms best” was con­duct­ed with them, ensur­ing that any mod­el that had a notice­able error rate was assessed for its impact in real scenarios.

Ulti­mate­ly, the mod­el was test­ed in prac­tice against man­u­al plan­ning by the oper­a­tors, who could still inter­vene in real time if any issues arose in the sup­ply chain. This approach not only max­imis­es the sup­port we pro­vide to the oper­a­tors and plan­ners but also acknowl­edges their years of valu­able experience.

Grif­fith Foods also played an instru­men­tal role in ensur­ing the suc­cess of this project beyond its ini­tial imple­men­ta­tion. Their lead­er­ship team active­ly col­lab­o­rat­ed with us to devel­op an inter­nal com­mu­ni­ca­tion strat­e­gy that high­light­ed the ben­e­fits and suc­cess­es of the cho­sen solu­tion. Recog­nis­ing the suc­cess achieved in their Bel­gian facil­i­ty, the lead­er­ship team at Grif­fith Foods is now explor­ing oppor­tu­ni­ties to repli­cate and adapt this mod­el in their pro­duc­tion sites in oth­er countries.

Engineer Adjusting Machine In Food Manufacturing Facility And Pointing Finger At Machine

Results

Less down­time due to wet clean­ings imme­di­ate­ly result­ed in a gain of 17 pro­duc­tion days. 5% few­er wet clean­ings are now need­ed thanks to the opti­mi­sa­tions of the Hype­r­i­on experts. 

The return on invest­ment for Grif­fith Foods is clear: 

  • Those 17 pro­duc­tion days per year that the machines can run extra, reveal a pay­back peri­od of only 10 months. 
  • To add to that, the project also had a pos­i­tive impact on sus­tain­abil­i­ty with­in Grif­fith Foods. Few­er wet clean­ings mean­ing a water sav­ing of 1000m² per year and less use of clean­ing agents.

Our services

Expert support when you need it most

Your busi­ness deserves more than just quick fix­es; it needs a part­ner who under­stands your goals and chal­lenges. With access to a net­work of over 500 experts, we’re equipped to pro­vide the right support.

Read more

Reliable managed IT services for business growth

Run­ning a busi­ness is chal­leng­ing enough with­out hav­ing to wor­ry about tech­nol­o­gy. With our tai­lored man­aged ser­vices, we keep your sys­tems secure, up-to-date, and run­ning smoothly.

Read more

Want to take the step together?

Reach out to us, and let’s explore how we can build future-proof solu­tions togeth­er. We’re just a mes­sage away!

Let’s sync up!

Send us a message