Coffee Application Intelligent Sensor Technology, Inc. 1
Customers More than 35 systems in the world. 3
Sensory test is difficult Market Research: We tried sensory evaluations, comparing to a competitor s products, but it was impossible to grasp market trends because there were too many samples. Sensory evaluation puts a burden on panelists. Results sometimes differ among panelists. New Product Development: It is difficult to communicate the actual taste of products, because there is no objective measurement of taste. Product Marketing: We have no confidence in whether the final prototype corresponds with development concept. It takes a lot of time and expense to provide market research for a new product. Quality Control: it is difficult to check the quality of many samples quickly and easily. An objective taste measurement is strongly required.
Low quality but inexpensive General classification of coffee High quality but expensive Canephora 35% Arabica 65% Temp., Time, etc. Roasting Blending Infinite combinations! It takes long time to become a coffee blender, because learning is accomplished through hands on training, and experience. By using the Taste Sensing System, Accumulated sensor outputs can provide objective information on their customers and consumers. Product design development time is greatly shortened, because the optimization of taste and the associated costs, cannot be achieved by a trained blender alone. K. Toko ed., Science and Technology to Produce Deliciousness of Foods/Medicines and Ensure The Safety, CMC Publishing (1) (in 5 Japanese).
POS(point of sales system) data with Taste data Regular coffee POS data, 3, yrs buyer taste preference Thin Thick Sour Bitter Sour Bitter Over Less than 6 Data provided by Taste Aroma laboratory inc.
Optimization calculation of coffee beans based on taste with market price Coffee bean importer, Ishimitsu co, Ltd Cost down more than 1% Reduction of stock more than 5% Data base of taste data of beans by TSS (1)Price + Taste data ()Copy taste of target product (3)Calculate blend ratio for target product at minimum cost Food Drug Technology to secure Taste and Safety Chapter 6 Development of Coffee product By use of Taste sensing system Author Ishimitcu Co., Ltd Mr. Ishiwaki 7
8
Example of Coffee product design Food design simulator Database Target product Compare by Taste sensor & Sensory test Bitterness 苦味 1-1 - Result Calculate the condition to find an optimum blend Cost Number of blendiing Quantity of coffee powder Taste importance Fine tuning by Taste panelist 酸味 コク キレ Sharpness 渋味 ターゲット商品 Target ブレンド試作品 Blend 1% 3% 5% Coffee Bean 1 Coffee bean Coffee bean Bitter Astringent Sour Sharp Price 苦味渋味酸味キレコク単価 苦味 Bitter Astringent 渋味酸味 Sour Sharp キレ コク Price 単価 3 1% 苦味 Bitter Astringent 渋味酸味 Sour Sharp キレ コク Price 単価 9
Example Evaluate Target product first and calculate the blending ratio 旨味コク 酸味 1 Bitter 苦味雑味 旨味コク 酸味 1 苦味雑味 Bitter TARGET -1 - -3-1 - -3 渋味 Astringen cy 旨味 渋味 Astringen cy 旨味 苦味 Bitter 塩味 ネスカフェコ ールト フ レント 1% 苦味 Bitter 塩味 UCC サ フ レント 11 53% 酸味 旨味コク 渋味 Astringen cy 酸味 1-1 - -3 苦味雑味 Bitter 旨味 旨味コク 渋味 Astringen cy 酸味 1-1 - -3 苦味雑味 Bitter 旨味 旨味コク Astringen 渋味 cy 1-1 - -3 苦味雑味 Bitter 旨味 苦味 Bitter ネスカフェ香味焙煎深煎 塩味 AGF マキシム 1% 苦味 塩味 36% Bitter Bitter 苦味目標品 Target 塩味ブレンド Blend 1
How to make coffee sample Pour 13g hot water Stir gently Weight 7 gram 1. Milling coffee bean. Take 7g sample 3. Put 13g hot water, keep 1sec. Stir slowly 5 times 5. Keep min 6. Filtering 7. Cool sample to ambient temp Filter Cooling 11
Taste feature from countries 1 酸味苦味雑味渋味刺激旨味塩味苦味渋味旨味コクモカ 酸味苦味雑味渋味刺激旨味塩味苦味渋味旨味コクガテマラ 酸味苦味雑味渋味刺激旨味塩味苦味渋味旨味コクコロンビア 酸味苦味雑味渋味刺激旨味塩味苦味渋味旨味コクキリマンジァロ 酸味苦味雑味渋味刺激旨味塩味苦味渋味旨味コクブラジル 酸味苦味雑味渋味刺激旨味塩味苦味渋味旨味コクマンデリン Bitter Bitter Bitter Bitter Bitter Bitter Bitter Bitter Bitter Bitter Bitter Bitter Brazil Columbia Guatemala Mocha Kilimanjaro Mandheling
Example: Optimum calculation for Target taste 旨味コク 苦味 Bitter マンデリン Mandheling 酸味 塩味 苦味雑味 Bitter 旨味 38.5% 渋味刺激 旨味コク 渋味 Bitter 苦味 キリマンジァロ Kilimanjaro 酸味 塩味.% Bitter 苦味雑味 旨味 渋味 渋味刺激 旨味コク 渋味 酸味 苦味雑味 Bitter 渋味刺激 旨味コク モカ Mocha 酸味 3.% 苦味雑味 Bitter 渋味 渋味刺激 旨味コク ブラジル Brazil 酸味 33.8% Bitter 苦味雑味 渋味 渋味刺激 苦味 Bitter 塩味 UCC Target リフレッシュ 旨味 マBlend 38.5%+キ.%+モ3.%+ブ33.8% 苦味 Bitter 旨味 Bitter 苦味 旨味 塩味 塩味 Confirm the taste of blended and target sample are same by taste panelist 13
Comparison between evaluated data and calculation data ブラジル マンデリン 旨味コク 渋味 酸味 苦味雑味 Bitter 渋味刺激 旨味コク 酸味 苦味雑味 Bitter 渋味 渋味刺激 苦味 Bitter 旨味 Calculate 苦味 Bitter 旨味 塩味 塩味 5% 5% 酸味 旨味コク 渋味 苦味雑味 Bitter 渋味刺激 マンデリン3.5g Actual + ブラジル3.5g 計算値 Estimate Bitter 苦味 旨味 塩味 Calculated value and actual measurement data are overlapped. 1
Comparison between TSS and sensory data(left is TSS Right is Nikkei MJ(13/3/) Sensory data by Nikkei TSS 各メーカー平均値 日本マクドナルド Sensory data by Nikkei TSS 各メーカー平均値 セブンイレブン 1円 1円 ローソン プロントコーポレーション 18円 円 ドトールコーヒー ファミリーマート 円 15円 ミニストップ 15円 サークルKサンクス 13円 15
Taste difference of Bean grade (Columbia and Brazil bean) Taste difference Ave. vs Columbia Taste difference Ave. vs Brazil Ave Excelso Suplemo Ave # #/5 16
出典 :http://www.keycoffee.co.jp/news/11/pdf/1118-1.pdf 17
Summary 1Copy target taste Blending ratio by Optimum calculation Blending study Foresee taste by changing ratio of blends 3Easy understanding of new product taste for buyers Propose space management by taste 5R&D by collaborating with POS data 18
Mathematical Optimization At first, installation of Excel add-in Solver is required. COFFEE LABORATORY WHITE STONE VA 578 TEL (8) 35-55
Step 1 Step Select a target product and measure it using the Taste Sensing System Enter the results into the Excel sheet Measure raw materials for product design using the Taste Sensing System Enter the results into the Excel sheet Step 3 Calculate the Mathematical Optimization Step Blend raw materials based on the result and perform fine tuning by a human sensory test if needed. COFFEE LABORATORY WHITE STONE VA 578 TEL (8) 35-55
Step1 Measure a target product using the Taste Sensing System. Enter the result into the Excel sheet. Enter the measurement result (estimated taste values) here. 3 COFFEE LABORATORY WHITE STONE VA 578 TEL (8) 35-55
Step Measure raw materials for product design. Enter the results into the Excel sheet. Enter the results (estimated taste values) here COFFEE LABORATORY WHITE STONE VA 578 TEL (8) 35-55
Step 3 1 Calculate the Mathematical Optimization: setting options Select the samples to be used for the optimization. If enter 1, the sample is to be used and not to be used in the optimization. Enter the minimum percentage of raw materials to be used in blending if needed. For example, if 3% is entered in the cell (E16), the raw materials contributed less than 3% will be automatically omitted from the result. As a result, the numbers of required raw materials for the final result will be decreases as shown in the cell (C16). COFFEE LABORATORY WHITE STONE VA 578 TEL (8) 35-55 5
Step 3 Calculate the Mathematical Optimization 1 1.Click the Data tab.click Solver 3.Click Solve 3 6 COFFEE LABORATORY WHITE STONE VA 578 TEL (8) 35-55
Step 3 3 Calculate the Mathematical Optimization.Check Keep Solver Solution and then click OK. 7 COFFEE LABORATORY WHITE STONE VA 578 TEL (8) 35-55
Step Blend the raw materials based on the calculated result and perform a human sensory test if needed. Optimized blending percentage 8 COFFEE LABORATORY WHITE STONE VA 578 TEL (8) 35-55