Instrumentation & Sensors for the Food Industry, Second Edition

Instrumentation and sensors for the food industry

Related titles from Woodhead’s food science, technology and nutrition list: Food chemical safety Volume 1: Contaminants (ISBN: 1 85573 462 1) This volume provides comprehensive information about contaminants in the food industry. The book opens with an explanation of risk analysis and analytical methods used for detecting contaminants in food products. This is followed by full details of relevant EU and USA regulations. The second part of the book provides information about specific contaminants. Food chemical safety Volume 2: Additives (ISBN: 1 85573 563 6) This volume provides comprehensive information about additives in the food industry. The book opens with an explanation of risk analysis and analytical methods in relation to the use of additives in food products. This is followed by full details of relevant EU and USA regulations. The second part of the book provides information about specific subjects including flavourings, sweeteners and colourings. Food process modelling (ISBN: 1 85573 565 2) A major trend within the food industry over the past decade has been the concern to measure, predict and control food processes more accurately in search for greater consistency, quality and safety in the final product. This book explores the current trends in modelling, their strengths and weaknesses and applications across the supply chain. It will be a valuable guide for production and technical managers within the food industry. Details of these books and a complete list of Woodhead’s food science, technology and nutrition titles can be obtained by: • visiting our web site at www.woodhead-publishing.com • contacting Customer services (e-mail: [email protected]; fax: +44 (0) 1223 893694; tel.: +44 (0) 1223 891358 ext.30; address: Woodhead Publishing Ltd, Abington Hall, Abington, Cambridge CB1 6AH, England)

If you would like to receive information on forthcoming titles in this area, please send your address details to: Francis Dodds (address, tel. and fax as above; e-mail: [email protected]). Please confirm which subject areas you are interested in.

Instrumentation and sensors for the food industry Second edition Edited by

Erika Kress-Rogers and Christopher J. B. Brimelow

Published by Woodhead Publishing Limited Abington Hall, Abington Cambridge CB1 6AH England www.woodhead.publishing.com Published in North and South America by CRC Press LLC 2000 Corporate Blvd, NW Boca Raton FL 33431 USA First published 2001, Woodhead Publishing Limited and CRC Press LLC ß 2001, Woodhead Publishing Limited The authors have asserted their moral rights. This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. Reasonable efforts have been made to publish reliable data and information, but the authors and the publishers cannot assume responsibility for the validity of all materials. Neither the authors nor the publishers, nor anyone else associated with this publication, shall be liable for any loss, damage or liability directly or indirectly caused or alleged to be caused by this book. Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming, and recording, or by any information storage or retrieval system, without prior permission in writing from the publishers. The consent of Woodhead Publishing Limited and CRC Press LLC does not extend to copying for general distribution, for promotion, for creating new works, or for resale. Specific permission must be obtained in writing from Woodhead Publishing Limited or CRC Press LLC for such copying. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation, without intent to infringe. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress. Woodhead Publishing Limited ISBN 1 85573 560 1 CRC Press ISBN 0-8493-1223-X CRC Press order number: WP1223 Cover design by The ColourStudio Project managed by Macfarlane Production Services, Markyate, Hertfordshire (e-mail: [email protected]) Typeset by MHL Typesetting Ltd, Coventry, Warwickshire Printed by TJ International, Padstow, Cornwall, England

Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . List of contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . List of symbols. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xi xvii xxvii

1 Instrumentation for food quality assurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Kress Rogers, ALSTOM, Ratingen 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Challenging conditions for sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Interpreting the readings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 Measurement types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5 Further reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix: Measurement types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 1 7 12 17 21 22 25

2 Instrumental measurements and sensory parameters . . . . . . . . . . . . . . . . . . . A. Hugi and E. Voirol, Nestle´ Research Centre, Lausanne 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 The nature of sensory perception . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Sensory evaluation methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Sensory-instrumental relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Summary and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

31 31 32 36 41 56 58 59

Part I In-line measurement for the control of food-processing operations . .

61

3 Principles of colour measurement for food . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. B. MacDougall, The University of Reading 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Colour vision: trichromatic detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

63 63 64

vi

Contents 3.3 3.4 3.5 3.6 3.7 3.8

Influence of ambient light and food structure . . . . . . . . . . . . . . . . . . . . . . . . Colour description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

66 68 72 74 81 82

4 Colour measurement of foods by colour reflectance . . . . . . . . . . . . . . . . . . . . C. J. B. Brimelow and P. Joshi, Nestle´ Research Centre, Lausanne 4.1 Introduction: food colour and quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Colour measurement principles: brief introduction . . . . . . . . . . . . . . . . . . . 4.3 Colour measurement methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Colour measurement of typical food materials . . . . . . . . . . . . . . . . . . . . . . . 4.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

85

5 Sorting by colour in the food industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J. M. Low, W. S. Maughan, S. C. Bee and M. J. Honeywood, Sortex Limited, London 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 What is a sorting machine? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Assessment of food particles for colour sorting . . . . . . . . . . . . . . . . . . . . . . 5.4 The optical inspection system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5 Completing the sorting system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6 Future trends: computer vision systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7 Using a colour sorter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.8 Further reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Food compositional analysis using near infra-red absorption technology I. B. Benson and J. W. F. Millard, NDC Infrared Engineering, Maldon 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Principles of measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3 Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4 Applications in the food industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5 The power of process monitoring and trending . . . . . . . . . . . . . . . . . . . . . . 6.6 Practical considerations for implementing on-line measurement . . . . . 6.7 Conclusions and the future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Practical aspects of infra-red remote thermometry . . . . . . . . . . . . . . . . . . . . . I. Ridley, Land Instruments International, Dronfield 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 Radiation thermometers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3 Measurement principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4 Practical situations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.5 Miscellaneous techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.6 Further reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

85 86 93 101 112 113 117 117 117 119 124 129 132 134 135 137 137 139 145 151 161 166 183 185 187 187 188 198 203 208 212

Contents 8 In-line and off-line FTIR measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Reh, Nestle´ Research Centre, Lausanne . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Food applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3 Calibration and general aspects of routine use . . . . . . . . . . . . . . . . . . . . . . . 8.4 Conclusions and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Microwave measurements of product variables . . . . . . . . . . . . . . . . . . . . . . . . . M. Kent, Kent and Partner Scientific Services, Biggar 9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2 Overview of microwave techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3 Dielectric properties and their parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4 Methods for measurement of dielectric properties . . . . . . . . . . . . . . . . . . . 9.5 Dielectric properties and measurement of bulk density and composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.6 Material structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.7 Apparatus for microwave measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.8 Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.9 Areas for development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.10 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix: Some manufacturers of microwave moisture measurement instruments suitable for foodstuffs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Pressure and temperature measurement in food process control . . . . . . P. G. Berrie, Endress+Hauser Process Solutions AG, Reinach 10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 Pressure measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.3 Temperature measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.4 General instrument design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.6 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Level and flow measurement in food process control . . . . . . . . . . . . . . . . . . . P. G. Berrie, Endress+Hauser Process Solutions AG, Reinach 11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Level measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3 Flow measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.4 Process automation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.6 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.7 User organisations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Ultrasonic instrumentation in the food industry . . . . . . . . . . . . . . . . . . . . . . . . N. Denbow, ND Technical Marketing, Alresford 12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2 Low-frequency techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.3 High-frequency techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.4 Contacts for further information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vii 213 213 219 227 228 229 233 233 234 235 245 251 260 261 268 272 273 277 280 280 281 287 295 299 302 303 303 303 310 316 322 323 325 326 326 327 336 352

viii

Contents 12.5 12.6

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13 Ultrasound propagation in foods and ambient gases: principles and applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Kress-Rogers, ALSTOM, Ratingen 13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2 Overview of ultrasound applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3 Speed of sound . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4 Acoustic impedance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.5 Attenuation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.7 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix: Ultrasound measurement applications in and for the food industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

353 353 355 355 356 366 383 385 393 394 400

14 In-line and on-line rheology measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. Roberts, Nestle´ Research Centre, Lausanne 14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.2 Requirements of an in-line or on-line sensor . . . . . . . . . . . . . . . . . . . . . . . . 14.3 In-line rheometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.4 In-line viscosity measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.5 Capillary (or tube) viscometers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.6 Rotational viscometers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.7 Vibrational viscometers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.8 High-frequency rheometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.10 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

403 404 405 407 408 413 416 418 419 419

Part II Instrumental techniques in the quality control laboratory . . . . . . . . . .

423

15 Rheological measurements of foods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. M. McKenna and J. G. Lyng, University College Dublin 15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.2 Relevance of rheological properties of foods . . . . . . . . . . . . . . . . . . . . . . . . 15.3 Basic rheology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.4 Measurement systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.5 On-line measurement systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.6 Instrument selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.7 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

425

16 Water activity and its measurement in food . . . . . . . . . . . . . . . . . . . . . . . . . . . . W. Ro¨del, Federal Centre for Meat Research, Kulmbach 16.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.2 Significance of water activity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.3 Water activity levels in food and their control . . . . . . . . . . . . . . . . . . . . . . . 16.4 Measuring water activity level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.5 Measurement techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

403

425 426 428 434 447 449 449 453 453 454 460 464 466

Contents

ix

16.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.7 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

474 474

17 Conductance/impedance techniques for microbial assay . . . . . . . . . . . . . . . D. M. Gibson, BIODON International, Aberdeen 17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.2 Rapid microbiological methods: an overview . . . . . . . . . . . . . . . . . . . . . . . . 17.3 Principles of electrical conductance methods . . . . . . . . . . . . . . . . . . . . . . . . 17.4 Capacitance versus conductance measurement . . . . . . . . . . . . . . . . . . . . . . . 17.5 Instrument design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.6 The evaluation of conductance data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.7 Future possibilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

484 484 485 489 495 496 503 510 512

18 Modern methods of texture measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Kilcast, Leatherhead Food Research Association 18.1 Introduction: texture and food quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.2 Perception of food texture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.3 Sensory assessment of texture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.4 Instrumental measurement of texture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.5 In vivo texture measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.6 Future developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

518 521 522 525 534 541 543 545

Part III Chemosensors, biosensors, immunosensors, electronic noses and tongues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

551

19 Sensors for food flavour and freshness: electronic noses, tongues and testers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Kress-Rogers, ALSTOM, Ratingen 19.1 Introduction to flavour assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.2 Modelling the human nose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.3 The electronic nose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.4 The electronic tongue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.5 The marker chemical approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.6 In situ freshness monitor for frying oil (resonant viscosity probe) . . . 19.7 Knife-type meat freshness tester (glucose profiling biosensor) . . . . . . . 19.8 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.9 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Chemosensors, biosensors, immunosensors and DNA probes: the base devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Kress-Rogers, ALSTOM, Ratingen 20.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.2 Chemically sensitive semiconductor devices: solid-state sensors for pH, acidity, ions, gases and volatiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.3 Amperometric, potentiometric and thermometric biosensors . . . . . . . . .

518

553 553 554 555 568 574 578 599 614 615 623 623 627 659

x

Contents 20.4 20.5 20.6 20.7

Chemically sensitive optical and acoustic devices . . . . . . . . . . . . . . . . . . . Applying sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Further reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

21 Biosensors for process monitoring and quality assurance in the food industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Schmidt and U. Bilitewski, National Research Centre for Biotechnology Ltd, Braunschweig 21.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Principles of immunoanalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.3 Detection of microorganisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.4 Microbial toxins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.5 Residue analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.7 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Commercial devices based on biosensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Warsinke, University of Potsdam, Golm and D. Pfeiffer and F. W. Scheller, BST Biosensor Technologie, Berlin 22.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.2 Principles of signal generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.3 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.4 Conclusions and future trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.5 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

675 688 693 694 714 714 715 722 728 729 732 733 733 740 740 740 747 756 756

23 New biosensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. Tothill, A. Piletsky, N. Magan and A. P. F. Turner, Cranfield University, Bedford 23.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.2 Novel sensing receptors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.3 Sensor arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.4 The electronic nose instrument . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.5 Genetically modified food . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.6 Commercial biosensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

760 760 761 766 766 768 769 771 771

Part IV Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A: Glossary: terms in instrumentation and sensors technology . . Appendix B: Ancillary tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

777 779 800

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

817

Preface

The optimization of industrial food processing The enjoyable, everyday food in an industrialized society relies not only on agriculture and fishing but also on efficient food processing. Few people mill their own cereals, preserve and store their fruit and vegetables from harvest to consumption, churn their butter, ferment milk into yoghurt and cereals into beer, turn meat carcasses into joints, sausages and paˆte´s, or pound mustard seeds to prepare meal accompaniments. Few wish to restrict themselves to the local products in season and most prefer instead the variety that modern food production, processing and distribution can offer. Many use readyprepared meals so as to spend their evenings with the children or with friends rather than in the kitchen. In the early stages of the industrialization of food processing, the competition between manufactured goods centred mainly on the price at which they were offered; now quality and safety are in the foreground. A wider range of attractive food products has become affordable for a large proportion of the population through advances in food science and technology together with the development of a diverse range of efficient large-scale processing plant. Many traditional batch processes have now been replaced by automated production methods, helped by the introduction of advanced process control systems in the 1980s. The signal processing and actuating capacity of process control systems is now adequate. The full potential of these systems, however, can be realized only if they are supplied with full and up-to-date information on the process to allow feedback or feedforward control. The development and knowledgeable application of sensors and instruments have become the key elements in meeting the consumer’s expectations in the food industry to provide affordable, enjoyable, safe and nutritious products. This has prompted the development of a wider range of sensors and instruments suitable for on-line and at-line measurements in the food industry, and also of modern instruments for the quality control (QC) laboratory. Many of the new instruments rely on a complex interaction with the food in order to determine properties of the food itself (such as composition) during processing. They extend the range of data inputs for

xii

Preface

‘sensible’ process control (equipped with senses) well beyond the measurement of pressure, temperature, level and flow rate. Other new instruments widen the range of applications for the measurement of the established variables, now allowing the reliable measurement of flow rate or temperature in food processes where this was previously impossible. Progress has also been made in the development of instruments for the assessment of food freshness and food safety, so that results are now often available within a day and a higher proportion of food ingredients and products can be screened to ensure good manufacturing practice. In the choice of instrumentation, an analysis of the processing operation as a whole, together with an overview of the characteristics of the sensors and instruments available for on-line, at-line and QC laboratory measurements, will be the basis of optimum process control design. On-line and off-line instrumentation interlink in guiding process control and are therefore both included here. Calibration samples need to be chosen and correctly prepared, and a representative sampling technique and suitable reference methods must be selected. For the reliable installation, calibration and operation of the new instruments, and for the correct interpretation of their readings, it is essential to understand the principles underlying the functioning of the instruments, the properties of the food and its processing environment, and their interplay. This approach also helps in assessing the many novel sensors and instrumental techniques now emerging to provide better longterm planning of process control optimization.

Special application details for instrumentation Instrument engineers coming from the aerospace, defence, nuclear or petrochemical sectors sometimes underestimate the challenges of designing sensors and instruments for the food industry. They find adequate challenges for their skills when they encounter a wide range of temperatures, pressures and pH values; mixing paddles continuously scraping container walls where a sensor is to be mounted in contact with the product; the rejection of guards around fragile sensor components as germ traps; and a limitation of the choice of engineering materials to those compatible with food hygiene considerations. The occasional fracture of a sensor in the chemical industry may be an inconvenience; in the food industry it is a major incident when any sharp fragments, however small, are lost into the process stream, requiring the screening or safe disposal of many thousands of food product items. A standard procedure for the maintenance of hygienic conditions in food processing is cleaning-in-place (CIP). This may sound harmless enough, but the periodic flushing of the food processing system with hot caustic soda (NaOH) solutions or pressurized steam places restrictions on the design of contacting sensors, particularly in the development of chemical sensors. Instruments based on non-contact methods are especially attractive to the food industry, being both intrinsically hygienic and easy to maintain. Such instruments are covered in the first part of this book. In some applications, hostile conditions and restricted access to the contents of a process vessel are the main challenges, for example in a cooker extruder which allows the continuous production of intricately shaped and textured snack foods at a throughput rate of 400 kg/hour. High pressures, high temperatures, a feed/mixer screw scraping the interior surface of the heavy metal barrel and sometimes abrasive raw materials combine here to render the construction of reliable sensors difficult, even for pressure and food mix temperature.

Preface

xiii

More often, it is the variable and complex nature of the food itself that presents problems in the design and application of instruments. This is the case for non-contact volume and mass flow rate metering of many foods. The most interesting problems arise, however, in the measurement of food properties such as composition or rheology. An interdisciplinary approach is needed here to take account of the interaction between the instrumental method and the chemical and physical properties of the food and its environment (beyond the variables to be determined). This applies also to the assessment of food freshness or conversely to the determination and prediction of changes due to microbial activity or oxidative processes. A further aspect is the perception of the consumer which needs to be represented in instruments for the assessment of appearance and texture. A rapid accurate measurement is often needed to maintain specifications within narrow margins. A pH value or water activity above specifications could lead to food spoilage during storage and distribution; a deviation to lower values could reduce the palatability. Too little preservative could endanger food safety; too much would be unacceptable to many consumers. Too high a water content could be infringing legal requirements or be associated with a water activity above specifications (with implications for food stability); too little water could result in an unattractive texture and an uncompetitive price for the food product. Line speeds in automated continuous food processing and packaging are high, and this is both a motivation for the application of on-line instrumentation (or of rapid at-line methods) and a challenge in the design of instruments for this purpose. A further constraint in the design of instrumentation for the food industry is the fact that the price of the sensor or instrument will be important in the purchasing decision. Whereas the aircraft constructor may well buy the best instruments at any price, the food industry cannot afford to do so.

Instrument types and aspects Instruments relying particularly on an interaction with the food or an environment typical for the food industry are described in this book. Practical applications already established are discussed and newly emerging applications are introduced. The considerations that will allow the best use of the interplay of the instrumental method, the food and the process are outlined as a basis for the successful development and implementation of instrument applications. Both on-line and QC laboratory instruments are included as they have to interlink in guiding process control. Instrument users often wonder why the flood of novel sensors and measurement techniques described in scientific and technical journals or at conference results in a mere trickle of novel commercial instruments. This has been the case particularly in the field of biosensors and chemical sensors based on microelectronic devices where rapid developments have taken place in recent years. Part III of the book illustrates the complex and expensive process of developing a novel instrument from concept to commercial fruition with the help of two examples. The basis of recent commercial instrument developments based on novel chemical sensors and the feasibility of further food applications are also examined there. For each instrument type, the underlying principles are described with emphasis on aspects relevant to food applications. The authors show the significance of the variables to be determined, and identify the variables actually measured (unless identical) and their

xiv

Preface

relation to the desired information about the food product. Considerations in the design, choice, calibration and running of instruments within a given group are discussed and illustrated with examples. Aspects covered include hygienic design (e.g. flush fitting sensor heads or choice of non-contact techniques) or the adaptation of techniques to the variable nature of food ingredients. (In three cases, two chapters deal with different aspects of the same technique.) Factors influencing the accuracy and reliability of the technique (for a particular group of food products if applicable) are spelled out and compared with alternative techniques where applicable. Instrument systems requiring a high computing capacity (such as real-time image acquisition and processing), employing ionizing radiation (such as gamma-ray density gauges) or relying on principles beyond the realms of classical physics are omitted to allow a full description of the instruments covered.

The authors’ background To promote an interdisciplinary understanding, these aspects are discussed here by scientists and engineers from a wide range of backgrounds including electronics, physics, chemistry, microbiology, food science and food technology. Their professional experience spans an equally wide range of areas within the fields of the development and application of instrumental methods for the food industry. The authors have worked in the management and optimization of quality control and quality assurance in the food industry, in the development of new techniques for this area, in applications development or overall management at an instrument supplier’s laboratory, or in a research institute or association in close contact with the food industry. It would be difficult to find a single author with full and detailed knowledge and practical experience in all the aspects of physics, electronics, chemistry, microbiology, food science, food technology and process control that are relevant to instrumentation in the food industry. Nor would it necessarily be helpful to set up a committee of authors to compose a text together. Instead, each chapter reflects the particular expertise of the author(s) based on their scientific or engineering background and their professional experience acquired in the practical application or development of instruments.

Aims and scope For a wide range of established and emerging instrument types, this book treats the underlying principles and their implications for industrial applications. It sets out the complementary roles and characteristics of both the on-line and at-line instrumentation linked to the process control system and of the off-line instruments in the quality control laboratory. The significance of the measured variables for quality assurance and process management and the technical and commercial factors that determine the success or failure of an instrument are considered. The book is intended to assist engineers and managers responsible for process optimization and quality assurance in the food industry in choosing, setting-up and maintaining instruments and in using their readings to best effect. It is also intended for use by engineers in the instrumentation sector who develop new instruments, adapt existing instruments for new applications or liaise with instrument users. In the choice

Preface

xv

and installation of an instrument in a process line, the effective cooperation between instrument supplier and user is essential and this book aims to promote this by facilitating the communication between engineers and managers, from different backgrounds. The chosen approach is also designed to help advanced students of instrument engineering, food science, physics or biochemistry who seek an introduction to instrumentation in the processing industries. Further, the book will be of interest to scientists active in research and pre-commercial development in the fields of process engineering, industrial instrumentation and process control. Erika Kress-Rogers and Chris J. B. Brimelow

Contributors

Chapters 1, 13, 19 and 20 Dr Erika Kress-Rogers (Alstom) Hamannstr 75 D-40882 Ratingen Germany Tel: +49 2102 51192 Fax: +49 2102 705204 E-mail: [email protected] Erika Kress-Rogers has a background in experimental solid-state physics (Universita¨t Karlsruhe) and in the physics of semiconductor devices (University of Oxford). She has carried out and coordinated a wide range of interdisciplinary R&D projects in the area of instrumentation and sensors for process control and quality assurance for the food industry while at the LFRA (Leatherhead Food Research Association), an international association that provides R&D, consultancy and technical services to food companies, government bodies and agencies. For seven years, she has served as Member of the International Editorial Board for the journal Food Control. She is the editor of the Handbook of Biosensors and Electronic Noses: Medicine, Food and the Environment and of the first edition of Instrumentation and Sensors for the Food Industry, the first handbook to provide a detailed account of a wide range of on-line and at-line measurement technologies for the determination of physical, chemical and microbial properties in the food industry. Dr Kress-Rogers now works as a Technical Editor for ALSTOM.

xviii

Contributors

Chapter 2 Dr Alain Hugi Nestle´ Research Centre Vers-chez-les-Blanc P.O. Box 44 CH-1000 Lausanne 26 Switzerland E-mail: [email protected] Dr Alain Hugi heads the Sensory Science Group in the Nestle´ Research Center (Lausanne – Switzerland). Prior to joining Nestle´ in 1996, he spent nine years at the Confectionery R&D Center of Kraft Jacobs Suchard, in Neuchaˆtel (Switzerland), occupying various positions ranging from analytical chemistry to sensory evaluation and to product development. His current interests include sensory texture assessment, the development of novel sensory and consumer testing tools, and sensory-instrumental relations. His academic background is in analytical chemistry (University of Lausanne) Elisabeth Voirol is a sensory analyst at the Nestle´ Research Center (Lausanne – Switzerland). In more than 20 years with Nestle´, she has worked on topics related to sensory perception, product optimisation, development of quality control methods and data acquisition systems. She has experience in a wide range of food products such as milk products, coffee, dehydrated food, pet-food, meat products, cereals, etc. More recently she has focused on sensory analysis of colour and the impact on consumer behaviour. Her background is in biology and neuro-physiology (Universite´ Pierre et Marie Curie, Paris)

Chapter 3 Dr Douglas B. MacDougall School of Food Biosciences University of Reading 4 Japonica Close Wokingham Berks RG41 4XJ England Tel: +44 (0)1189 780174 E-mail: [email protected] Dr Douglas B. MacDougall lectured in sensory analysis at Reading University until his retirement in 2000. His research at Reading concentrated on the use of sensory analysis techniques to quantify food quality and the relationship of instrumental methods of colour measurement with the visual colour/appearance of food. Earlier, he worked in the Consumer Science Section at the AFRC (Agricultural and Food Research Council) in Reading and before that as co-project leader and Head of the Colour Group at the AFRC Meat Research Institute in Langford, Bristol. Projects at the Institute of Food Research included studies of the effects of animal stress and meat processing on the colour appearance, translucency and colour stability of fresh meat; optical instrument development for detection of meat faults; mathematical modelling of the colour and

Contributors

xix

texture of composite meat products and studies on the visual appeal of meat as affected by processing conditions, storage and display illumination. His background is in food science; colour stability of food (Royal College of Science and Technology, now Strathclyde University, Glasgow, Scotland; Rutgers, The State University of New Jersey, New Brunswick, USA).

Chapter 4 Dr Pallavi Joshi Nestle´ Research Centre Vers-chez-les-Blanc Case Postale 44 CH-1000 Lausanne 26 Switzerland Tel: +41 21 785 8540 Fax: +41 21 785 8554 E-mail: [email protected] Pallavi Joshi is a research scientist in the Department of Quality & Safety Assurance, Nestle´ Research Center for Food and Life Sciences, Lausanne Switzerland. Her current research focuses on the use of colour physics and measurement as a tool for product development and quality control within the food industry. Christopher J. B. Brimelow is head of Nestle´ R&D Centre Shanghai Ltd. He was formerly Vice President of R&D at Nestle´/Westreco (Connecticut – USA). He has worked on the on-line and off-line measurement of the compositional and physical properties of foods, particularly colour.

Chapter 5 Dr Sarah Bee Research Co-ordinator R&D Department Sortex Ltd. Pudding Mill Lane London E15 2PJ England Tel: +44 (0)20 8522 5136 Fax: +44 (0)20 8519 3232 E-mail: [email protected] Sarah C. Bee is Research Co-ordinator for Sortex Limited (London, UK). Sarah has worked in the R&D department for over three years, initially providing technical support for R&D, customer care, applications, production and sales and marketing. She currently initiates and subsequently manages Sortex’s external research interests, either with universities or commercial consultancies. Sarah has a background in radiation physics (University College London), is a Chartered Physicist and Member of the Institute of Physics.

xx

Contributors

Mark J. Honeywood is Sortex’s Technical Director. Through Sortex’s membership of the DTI’s Insider UK Enterprise (IUKE) scheme, Mark has been invited to present at the DTI Successful Product Development Seminars, held across the UK. Mark has first hand experience of implementing ‘fast cycle time’ techniques, including rapid prototyping, concurrent engineering and quality functional deployment. His background is in applied optics (Reading University). Mark is also a Chartered Physicist and a Member of the Institute of Physics.

Chapter 6 Dr Ian B. Benson NDC Infrared Engineering Maldon Essex CM9 4XD England Tel: +44 (0)1621 852244 Fax: +44 (0)1621 856180 E-mail: [email protected] Dr Ian B. Benson joined Ilford films in 1978 as a photographic research chemist. After a period in product development he joined Infrared Engineering as an applications engineer in 1981. After establishing laboratory and development facilities in the company’s head office in Maldon and developing a wide range of measurements he moved into the sales management role and is now Director of Marketing for the Instrument Gauging Business. James Millard studied applied physics at Coventry Lancaster Polytechnic, graduating with an honours degree. Having spent an industrial training year at Infrared Engineering he returned as an applications engineer in 1987 and was seconded to the USA operation in 1990. After managing the Technical Support Group for the global business he has now become a Product Manager for the instrument gauging business.

Chapter 7 Mr Ian Ridley Land Instruments International Limited Dronfield S18 1DJ England Tel: +44 (0)1246 417691 Fax: +44 (0)1246 410585 E-mail: [email protected] Ian Ridley studied applied physics at Sheffield City Polytechnic and since starting with Land Infrared in 1978 has been involved in the design and development of a wide variety of infrared-based temperature measurement equipment. He is now the Products Group Manager within the New Developments and Applications Department of Land Instruments International Ltd.

Contributors

xxi

Chapter 8 Dr Ing. Christoph Reh Nestle´ Research Centre Nestec Ltd Vers-chez-les-Blanc 1000 Lausanne 26 Switzerland Tel: +41 21 785 8990 Fax: +41 21 785 8553 E-mail: [email protected] Dr Ing. Christoph Reh works within the Nestle´ Research Center (Lausanne – Switzerland).

Chapter 9 Dr Michael Kent Kent and Partners Scientific Services 162 High Street Biggar ML12 6DH Scotland Tel: +44 (0)1899 220305 Fax: +44 (0)1899 220305 E-mail: [email protected] Mike Kent runs his own consultancy, Kent and Partner. He was formerly Head of Physics Section at the Torry Research Station, Aberdeen, where he carried out work on the dielectric properties of foods and the applications of such properties to compositional measurement.

Chapters 10 and 11 Dr Peter G. Berrie Endress+Hauser Process Solutions AG Christoph-Merian-Ring 23 4153 Reinach BL Switzerland Tel: +41 61715 7340 Fax: +41 61715 7301 E-mail: [email protected] Dr Peter G. Berrie works as Marketing Communications Manager for Endress+Hauser Process Solutions AG, Reinach, Switzerland. A graduate of Imperial College, London, he spent five years in research at Euratom in Karlsruhe, Germany and Loughborough

xxii

Contributors

University, England, before turning to technical communication in 1978. Dr Berrie arrived at Endress+Hauser GmbH+Co, Maulburg, Germany in 1990, as a technical author responsible for digital communication, level and pressure products. In 2000, he moved to his current position, where he is concerned with fieldbus technologies and process solutions that include sensors, monitoring and control. Endress+Hauser has an international reputation in the field of process instrumentation for the food and beverages industries.

Chapter 12 Mr Nicholas Denbow ND Technical Marketing 7 Carisbrooke Close Alresford Hants SO24 9PQ England E-mail: [email protected] www.nickdenbow.com Nicholas J. Denbow is currently a self-employed consultant in industrial instrumentation, specialising in ultrasonic techniques particularly for liquid level and fluid flow measurement. Previously employed as Marketing Manager for Platon Instrumentation at Basingstoke and Technical Marketing Manager for Solartron Mobrey in Slough, he has worked in industrial instrumentation for the process industries for 25 years.

Chapter 14 Dr Ian Roberts R&D/ QS Nestle´ Research Centre Nestec Ltd. Vers-Chez-Les-Blanc 1000 Lausanne 26 Switzerland Tel: +41 21 785 8469 Fax: +41 21 785 8553 E-mail: [email protected] Ian Roberts performed his undergraduate and postgraduate studies in the department of Chemical and Biochemical Engineering at the University of Wales, Swansea. Here, he obtained Bachelors and Masters degrees, before focussing on rheology in his PhD entitled ‘Rheometry for Gelling Systems’. Having performed his thesis in collaboration with Nestle´ UK, he then moved to the Nestle´ Research Center in Lausanne, Switzerland in 1997, and now works at the Nestle´ Product Technology Centre in Orbe, Switzerland.

Contributors

xxiii

Chapter 15 Dr James G. Lyng Department of Food Science University College Dublin Belfield Dublin 4 Ireland Tel: +353 (0)1 7067710 Fax: +353 (0)1 7061147 E-mail: [email protected] Professor Brian M. McKenna is the Head of the Food Science Department at University College Dublin (UCD), and is also the Director of the Food Science Centre at UCD, in addition to being Vice President of UCD and editor of the Journal of Food Engineering. Professor McKenna lectures in physical properties of foods at UCD. His research interests include the freezing and drying of foods, membrane processing, cook-chill products, process modelling, shelf-life prediction and electroheating. Professor McKenna has experience in the measurement of rheological and many other physical properties of foods through his involvement in national and EU-funded food process technology and product property research projects. Dr James Lyng is a lecturer in the Department of Food Science at UCD. In addition to giving courses in food process technology and food engineering, he also lectures in physical properties of food with Professor McKenna. His research interests lie in the area of alternative processing systems for meat and meat products (with particular reference to electroheating) and a large proportion of this work involves the measurement of physical properties of these products particularly their thermal, dielectric and also textural and rheological properties.

Chapter 16 Dr Wolfgang Ro¨edel Director and Professor Federal Centre for Meat Research Amselweg 16 D-95326 Kulmbach Germany E-mail: [email protected] Wolfgang Ro¨del is Director and Professor at the Federal Centre for Meat Research and Vice Head of the Institute for Microbiology and Toxicology (Kulmbach, Germany). His research interests include development and adaptation of electronic measurement procedures for the determination of the physicochemical parameters (water activity, redox potential, pH, etc.) of meat and meat products within the framework of HACCP and quality control.

xxiv

Contributors

Chapter 17 Dr Donald M. Gibson BIODON International 43 Brighton Place Aberdeen AB10 6RT Scotland Tel/fax: +44 (0)1224 322 777 E-mail: [email protected] Donald M. Gibson is an independent consultant with his own company, BIODON International, established in 1994. He specialises in food microbiology and technology. He spent 30 years at the Torry Research Station, Aberdeen, latterly as Head of Microbiology, covering many aspects of microbiological food safety and quality. He is a Fellow of the Institute of Food Science and Technology, and a member of the Society for Applied Microbiology and of the Association of Official Analytical Chemists International (AOAC).

Chapter 18 Dr David Kilcast Leatherhead Food Research Association Randalls Road Leatherhead Surrey KT22 7RY England Tel: +44 (0)1372 822321 Fax: +44 (0)1372 836228 E-mail: [email protected] David Kilcast, BSc, PhD, FIFST is Head of Sensory and Consumer Science at Leatherhead Food Research Association and leads a research team working on the sensory quality and consumer perception of food. Research specialities are the perception and measurement of flavour and texture including flavour release from foods. He is past Chairman of the Sensory and Consumer Science Group of the Society of Chemical Industry, a member of the British Standards Institution Committee on Sensory Analysis and a committee member of the IFST Professional Food Sensory Interest Group.

Chapter 21 Dr Ursula Bilitewski GBF-Ges. Biotechn. Forschung mbH Mascheroder Weg 1 D-38124 Braunschweig Germany Tel: +49 531 6181 390 Fax: +49 531 6181 395 E-mail: [email protected]

Contributors xxv Dr Ursula Bilitewski is senior scientist in the German Research Centre for Biotechnology (GBF), Braunschweig, Germany and lecturer in biochemistry at the Technical University Braunschweig. In the Division of Biochemical Engineering of the GBF she is responsible for the development and application of bioanalytical methods. She has a long experience with electrochemical methods and has used screen-printing technology for the production of enzyme electrodes to be applied in food and bioprocess analysis. There are also strong activities in the design of automated flowthrough devices, which were used not only for enzyme, but also for immunoanalysis, and included electrochemical as well as optical detection methods. Recent research activities cover the analysis of proteins and protein activities, the analysis of genes and gene expression in combination with the miniaturization of set-ups. Dr Anja Schmidt is a food chemist and was a PhD student and postdoc in Dr Bilitewski’s group.

Chapter 22 Dr Axel Warsinke Department of Analytical Biochemistry Karl-Liebknecht-Str. 24-25 D-14476 Golm Germany Tel: +49 331 977 5124 Fax: +49 331 977 5052 E-mail: [email protected] Dr Axel Warsinke works within the Department of Analytical Chemistry at the University of Potsdam. Professor Dorothea Pfieffer and Dr Frieder Scheller work for BST Bio Sensor Technologie GmbH based in Berlin.

Chapter 23 Dr Ibitsam Tothill Institute of Bioscience and Technology Cranfield Biotechnology Centre Cranfield University Cranfield Bedfordshire MK43 0AL England Tel: +44 (0)1234 754131 Fax: +44 (0)1234 752401 E-mail: [email protected] Dr Ibtisam E. Tothill is a senior lecturer in biochemistry and MSc Course Director for the MSc in Environmental Diagnostics at Cranfield Biotechnology Centre, Cranfield

xxvi

Contributors

University, Bedfordshire. She has developed strong research activities in immunosensors, affinity sensors, biosensors and diagnostics, which are at the forefront of research into food and environmental analysis. She has established an international and European reputation in these areas and this has enabled her to attract both private and public sector funding. Her recent research activities cover analysis of pesticides and herbicides, algae and cyanobacterial toxins diagnosis, heavy metals detection, bacteria and fungal detection. Dr Tothill has numerous publications and conference papers in the diagnostics and analysis arena. She has received a variety of prizes and awards including the Douglas Bomford Trust Award at the AgEng 2000, Warwick, UK.

Symbols

Chapter 13 c  Z  T f  ! k   P M  rp V T EM K Y G KS P

cP/cV  R B STP

speed of sound dispersive ultrasound phase velocity specific acoustic impedance coefficient of ultrasound attenuation coefficient of transmission through boundary frequency wavelength angular frequency; ! ˆ 2f wave vector; k ˆ 2= relaxation time density absolute pressure molecular weight coefficient of viscosity radius of suspended spherical particle volume absolute temperature elastic modulus (as appropriate) bulk modulus Young’s modulus shear modulus adiabatic bulk modulus coefficient of expansion at constan pressure ratio of principal specific heats cP/cV specific heats at constant pressure, volume coefficient of thermal conductivity universal gas constant second virial coefficients standard temperature and pressure

xxviii

Symbols

Chapter 15 An Bn d e f G G0 G00 g H h hc h0 I2 K Kn k k0 k 00 L Lo L0o n
p Q Rc R1 R2 Rp
R T Te t t0 up V 

_

_ eff

_ max

_ meas   0

function in equation (15.22) function in equation (15.22) capillary tube diameter distance from bottom of bob to cup or container frequency shear modulus storage modulus loss modulus gravitational constant separation of the two plates difference in levels between liquid in reservoirs of a capillary viscometer end effects correction in concentric cylinder viscometry height of bob for a rotary viscometer Bessel function constant for capillary viscometer root of modifier Navier-Stokes equation apparent viscosity or consistency index for power law fluid constant in Casson equation constant in Herschel-Buckley length of capillary tube sample length before deformation final sample length after deformation power law exponent pressure drop along a capillary tube flow rate radius of cone radius of bob in concentric cylinder viscometer radius of cup in concentric cylinder viscometer radius of the plate (parallel plate systems) R2 R1 torque equivalent torque time (equations (15.8) (15.9) or (15.12)) relaxation time (equation (15.4)) slope of plot of shear stress versus shear rate once yield stress has been exceeded volume of liquid used in a capillary viscometer conical angle relaxation time shear strain …ˆ …L0o Lo †=Lo , dimensionless) shear rate (or rate of shear strain) d /dt effective shear rate shear rate at the outer edge (rim) of the plate measured shear rate phase displacement angle viscosity viscosity at zero shear rate

Symbols p 1 !    b y

xxix

slope of straight line plot of shear stress vs. shear rate (equation 15.17) viscosity at infinite shear rate angular velocity 3.1417 density shear stress shear stress at the bob yield stress

Chapter 17 b c f k l n n0 ns nD n_ _ n/n _ 0 (n/n) r t tg tg0 tD tL A AD CD, R D Cox, R ox Cse G0 Gs K N R fs Rs S S0, S s
S

number of bits mass concentration of medium, g ml 1 frequency, Hz initial specific conductivity, S m 1 effective separation of electrodes, m population density, cfu ml 1 inoculum density, cfu ml 1 maximum possible population density, cfu ml 1 arbitrary detection level of population density, cfu ml 1 growth rate, cfu h 1 specific growth rate, h 1 initial specific growth rate, h 1 correlation coefficient time, h generation time, h _ o, h 1/(n/n) time after inoculation at which population density grows to a value nD, detection time, h lag time, h electrode area, m2 conversion gain series capacitance and series resistance due to alignment of polar dipoles in double charge layer in fluid, F,

series capacitance and series resistance resulting from the presence of an oxide layer at the electrode surfaces, F,

lumped electrode capacitances CD and Cox (for R D, R ox  R s), F initial conductance of medium in test cell, S subsequent conductance of medium in test cell, S increase in specific conductance associated with production of single bacterial cell, per unit volume, S m 1 (cfu ml 1) 1 number of colonies counted on agar plate full-scale range of resistance,

true resistance of medium in test cell,

substrate mass, g initial and final stationary mass of substrate, g mass of substrate metabolised by one bacterial cell, g

xxx

Symbols

V, i |V|,|i| |Vm|,|im| Xs Xse Y Z 

D

instantaneous voltage, current, V, A peak voltage, current, V, A mean voltage, current, modulus, V, A reactance of Cs,

reactance of Cse,

admittance, S impedance of R s and Cs in series,

phase angle number of generations after inoculation at which growth is detected

1 Instrumentation for food quality assurance E. Kress-Rogers, ALSTOM, Ratingen

1.1

Introduction

1.1.1 The role of quality assurance in the food industry Quality control is essential in the food industry, and efficient quality assurance is becoming increasingly important. The consumer expects a wide range of competitively priced food products of consistently high quality. Each food item has to be safe, wholesome and attractive in appearance, taste and texture, and needs to be consistent with the product image. Variations within the same batch or between batches will have to be kept to a minimum as they are often interpreted by the consumer as indicating a fault, even when the differing product is of high quality. The availability, quality and price of raw materials will place conditions on the food manufacturing operation, as will the prevailing structure of the retailing sector. More and more frequently, the product palette has to be adapted to changes in tastes and nutritional ideas, and to the appearance of competing products on the market. In the manufacture of each new product, there is the challenge of getting it right first time. Increasingly, food processing operations are technology-based rather than skill-based. Legislation on food composition and labelling will also play a role. Changes in legislation are driven by consumer demands and by international harmonization. Food processing has a long history (Georgala 1989) and has always had two main purposes. The first is the conversion of agricultural products (or of fished, hunted and gathered foods) into palatable, attractive, digestible and safe foods. Cereals, for example, are virtually inedible without prior milling and cooking or fermenting; some fruits and pulses are toxic without prior cooking; and large proportions of the Asian and African populations can consume lactose only after conversion to lactate by fermentation. The second purpose is the preservation of foods for availability out of season, for years of lean harvests, and for transport to areas distant from agricultural producers. The assessment of food still centres on its taste, aroma, appearance and nutritional value, and on its safety and stability. Optimized process control plays an essential part in maintaining the commercial viability of a food manufacturing operation in the face of changes in the food market and in the structure of the food industry. Advances in

2

Instrumentation and sensors for the food industry

Table 1.1

Food and drink market sectors in the UK (Bailey et al. 1991)

(a) Market sectors in 1990 by value (£ million) Alcoholic beverages (inc. duty)* 1 Meat and meat products 2 Dairy products (inc. ice cream) 3 Soft drinks 4 Fresh fruit and vegetables 5 Bakery products 6 Confectionery 7 Frozen foods 8 Canned foods 9 Fish and fish products 10 Snack foods 11 Hot beverages 12 Cereal products 13 Oils and fats products 14 Ready meals 15 Meal accompaniments 16 Sweeteners, preserves

21 9 5 5 4 3 3 2 1 1 1 1 1

864 882 884 286 598 790 700 250 686 606 287 281 230 971 715 678 535

(b) Increase of market sector values in the UK from 1989 to 1990 (per cent) Alcoholic beverages* 1 Frozen foods 2 Soft drinks 3 Ready meals 4 Confectionery 5 Fresh fruit and vegetables 6 Snack foods 7 Cereal products 8 Meal accompaniments 9 Hot beverages 10 Dairy products 11 Meat and meat products 12 Fish and fish products 13 Bakery products 14 Oils and fats products 15 Canned foods 16 Sweeteners, preserves

8.2 11.2 10.9 10.5 8.8 8.2 6.5 5.4 4.3 4.0 3.5 3.5 3.3 2.1 1.4 1.1 0.6

Notes: * As much of the market value of the alcoholic beverages sector is the duty, the value has not been used to rank this sector. 1. For the cost/benefit assessment for a potential instrument development, the market values listed here need to be seen together with other factors such as the relative values of raw materials and final product or the growth rate and profit margins in a sector. For snack foods, for example, the added value would, in general, be higher than for canned foods or meat. Additionally, the viability of an instrument development is increased when it is relevant to health and safety (where legislation is linked to the availability of instruments) or to the price of commodities (for example, in the case of the water content of wheat) or where the specifications for the proportional content of an expensive ingredient need to be met. 2. All market sectors increased in value, but not all growth rates exceeded the rate of inflation. 3. Categories are not mutually exclusive.

Instrumentation for food quality assurance

3

microelectronics have provided fast data processing and have made efficient process control systems possible. In the 1980s, programmable logic controllers (PLCs) were widely installed in the food industry. Massive control centres were designed earlier for integral plant control; these centres were subsequently replaced by distributed control systems (McFarlane 1983; Vidal 1988). Whichever control system is used, it still has to make do with a small number of continuously updated product variables, and often relies largely on inputs at long time intervals and with long delays depending on the assay time and the distance to the quality control (QC) laboratory. The effective application of both established and novel sensors and instruments will play a key role in gaining the full benefit of the potential that modern control systems offer. Table 1.1 shows the sizes and growth rates of food and drink market sectors in the UK. 1.1.2 On-line, at-line and off-line instrumentation For optimum quality assurance the manufacturer requires cost-effective methods for the rapid assessment, and preferably the on-line measurement, of the chemical and physical properties and the microbial status of raw materials, process streams and end products. Monitoring during the processing operation helps prevent expensive rework or disposal of out-of-specification product. Tight control is needed for variables that influence the stability of the end product towards microbial spoilage or oxidative rancidity. This concerns particularly the monitoring of temperature profiles during heat processing and storage, the control of cleaning-in-place procedures, and the measurement of the pH, water activity, solute concentration and preservative levels of the product. Water activity, usually measured as equilibrium relative humidity (ERH), cannot be measured rapidly. From an on-line measurement of the moisture content, the ERH can be deduced if the isotherm is well defined. The trend towards continuous automated production in place of batch processing necessitates tight feedback loops based on on-line monitoring methods or, failing that, on rapid at-line techniques. Even when a laboratory method provides a result within one hour of taking a sample from the line, over a tonne of product or over 10 000 jars, tins or packs of food may already have passed the production line. The cost of rework or disposal for such a quantity is considerable. Alternatively, excessive safety margins with respect to legal requirements or customer specifications on the minimum content of expensive ingredients will lead to an uncompetitively priced product. Prolonged holding times to await the outcome of assays, as a regular part of the process, lessen the benefits of continuous processing. Nevertheless, holding times of around eight hours are currently observed prior to filling certain sterilized foods, for example, in order to await test results from impedance monitoring for microbial assessment (Chapter 17). Refinements of this technique, based on more sensitive oscillometric detection of impedance changes with microbial growth, for example, have been investigated in order to shorten the assay time (Cossar et al. 1990). The advances in plant for automated continuous production and in the signal processing capabilities of process control systems have stimulated progress in the development of many novel sensors and instruments for the food industry, often by technology transfer from other industrial sectors or from the clinical sector (Kress-Rogers 1985, 1986). These have since matured; sensor concepts have been developed into prototypes, and instrument types already available in the 1980s have become more

4

Instrumentation and sensors for the food industry

versatile and can now be applied reliably to a wider range of foods and processing situations or determine a wider range of target variables. With the help of these advances in on-line and at-line instrumentation (Fig. 1.1), quality assurance (QA) is employed increasingly in the management of manufacturing operations. The quality control (QC) laboratory supports QA by checking and updating the calibration of on-line and at-line instrumentation and by providing a wide range of analyses and assessments that are not feasible for QA implementation. The variables measured on-line and those measured off-line in the QC laboratory do not necessarily coincide. The process stream at the on-line measurement point will often be quite different from the sample taken to the laboratory, either due to changes during sampling and transporting, or because the laboratory test measures properties of the end product, whereas the on-line instrument measures precursors of these, or other properties of the process stream or the process environment that will determine the relevant properties of the product. When the time taken for a QC laboratory result exceeds a day, as would be the case for many microbiological tests or trace analysis assays for toxins, it is often impractical to hold the food in quarantine during this time, as a perishable food may be well into its shelf-life by the time the result is available. Even when prolonged holding times can be observed, it is not usually possible to provide 100 per cent screening of the product with QC methods, and so a negative result is no absolute guarantee that the whole production volume is ‘clean’. The test then becomes a means of checking that good manufacturing practice (GMP) is being observed, and the process has to be analysed to define the product and process variables that can be monitored and controlled in order to minimize the possibility of manufacturing a product having too high a microbial load, carrying pathogens or containing toxins. This approach is known as hazard analysis critical control point (HACCP) system. For the overall control of the process, the monitoring of level and flow rate as well as pressure and temperature are essential (Chapters 10–13). Important for the stability of foods towards microbial spoilage are product properties such as the water activity and the pH as well as the microbial load and the concentration of preservatives and nutrients (Chapters 16, 17, 20–23). The integrity of the food packaging is also vital, and in modern modified atmosphere packs (MAPs) the initial headspace gas composition and its retention during distribution and storage will be relevant. The adherence to appropriate storage temperatures (and ambient humidities) throughout the shelf-life needs to be ensured. An important process variable influencing the shelf-life is the time-temperature profile of the process stream and, related to this, the excess pressure in the headspace. Also relevant are the concentration and temperature of cleaning liquids and their efficient application to process plant surfaces (Chapters 10, 12, 13). In conventional cooking and canning operations, heating the interior of a solid food item (or a highly viscous liquid) relies on thermal conduction, often resulting in overcooking of the outer layers in order to ensure adequate temperatures in the centre. This is not to say that high surface temperatures are not desirable in processes such as roasting, where the Maillard reaction provides a range of flavours and colours in the presence of reducing sugars and amino acids at elevated temperatures. The flavour changes caused by prolonged boiling are, however, usually considered undesirable. Microwave or radiofrequency waves, on the other hand, can penetrate food and heat deeper layers directly. Direct ohmic heating is also possible by mounting electrodes in contact with a conductive food, and ohmic heaters allowing continuous automated heat processing are available.

Instrumentation for food quality assurance

5

With these methods, it is possible to retain more of the flavour and vitamins of the food, and yet to ensure a given minimum temperature to be reached throughout. In order to optimize such processes for the manufacture of products that combine adequate cooking, pasteurization or sterilization (as required for the product) with good flavour retention, analysis of the spatial distribution of the time-temperature profiles is necessary. Several variants of time-temperature integration are used to assess the effect of heat processing on a food. The most common is the F0 value, which expresses the degree of sterilization of a food. The F0 value (expressed in minutes) is obtained by calculating the integral Z F0 ˆ L dt where lg L ˆ (T Tref)/Z defines the lethality L, and the temperature T has been measured in the coldest part of the food. For F0 evaluation, Tref ˆ 121ºC and Z ˆ 10. For canned foods and ultra heat treated (UHT) products, F0 values of 3 to 18 are used, depending on the types and numbers of spores present. This treatment results in commercial sterility, that is the remaining microorganisms will not cause spoilage or disease or have a detrimental effect on the product quality during its stated shelf-life (usually in excess of six months) (Lewis 1987). Other values of Tref and Z apply for the loss of nutrients by protein denaturation, vitamin destruction and certain other chemical reactions. A cook value can be defined, in analogy to the sterilization value F0, to quantify the degree of cooking or overcooking and thus predict the loss of quality (flavour, nutrient levels) by heat processing. Figure 1.2 summarizes the roles of on-line, at-line and off-line instrumentation in process management, quality assurance and quality control. Measurements relevant for product safety, stability and quality and for process management are listed in Tables 1.2 and 1.3. Instrument requirements and measurements for special concepts are given in Tables 1.4 and 1.5. 1.1.3 Technology transfer: opportunities and pitfalls Instruments for measurements in quality control and in the control of processing operations in the food industry are often the result of technology transfer from other industries. The history of such new introductions has, in some cases, been characterized by initial successes, followed by a phase of disappointment with the instrument performance when the range of applications was widened. Subsequently, lost confidence had to be regained by defining the range of suitable application areas and by adapting the instrument or the setting-up and running procedures to particular applications. To avoid setbacks, it is necessary to understand both the instrument design and its underlying principles as well as the properties of the food and its processing environment. Problems have, for instance, been experienced with some early applications of ultrasound flow meters in the food industry. These flow meters have the attraction of providing a non-contact measurement which facilitates maintenance and is intrinsically hygienic. However, for certain food process streams, unacceptable errors in the readings were observed until it was recognized that special designs or other types of flow meters were needed for samples with non-Newtonian flow profiles, or containing large particulates with flow rates differing from that of the carrier liquid, or where high attenuation of the ultrasound signal by the food liquid restricted the sampled flow volume to the outer layer (Chapters 12, 13, 15).

6

Instrumentation and sensors for the food industry

Fig. 1.1 Sensor configurations: (a) sensors on continuous processing lines (b) sensors on conveyor belts (c) sensors in batch processes (d) handheld sensors. The window material will depend on the instrument principle, for example Teflon for microwave transmission. Conditions for at-line measurements (not on-line, but in the production area) are more stringent than for off-line measurements (in the QC laboratory). At-line instrumentation and accessories should be free of glass components (potential foreign body hazard) and of toxic reagents that are not fully contained at all times. Also, mechanical robustness, tolerance of the processing environment (for example, of steam) and simple and rapid operation are essential.

Instrumentation for food quality assurance

1.2

7

Challenging conditions for sensors

1.2.1 Complex and variable samples Many foods are highly complex in their chemical composition and in their physical structure. Gaseous, liquid and solid phases often coexist in the same product. Each phase may incorporate many different chemical compounds. One phase can be finely dispersed in another, or samples can be highly inhomogeneous or even largely separated. Within the

8

Instrumentation and sensors for the food industry

Fig. 1.2

Quality assurance (QA) and quality control (QC) in food processing operations.

Actuation of process changes A programmable logic controller (PLC) may be used to operate actuators that cause changes in process conditions in dependence of a measured variable. Where complex relationships exist between the measured variables and the process, an expert system can provide an automatic evaluation of a set of measured values and a decision on corrections to the process conditions. Combined inspection/sorting systems are used to identify and remove, for example, products that contain foreign bodies or that are mis-shaped. In-situ measurements in batch processes The scheme needs to be adapted for industrial batch processing operations, for batch processes in catering establishments and for measurements in food distribution. In-situ measurements with dip- or stab-probes, or with instruments installed during the batch process or permanently in a processing vessel or storage container can be used here.

Calibration The choice of the reference method can influence the calibration. Systematically different values can be obtained, for example, between drying and titration methods for moisture determination in the laboratory.

Instrumentation for food quality assurance Table 1.2

9

Measurements in quality assurance and quality control

Measuring properties relevant for product quality • • • • • •

appearance (colour, gloss, shape) texture, mouthfeel, pouring characteristics flavour (aroma, taste) nutritional value functional properties composition according to specifications

Screening for product safety chemical contamination (agricultural residues, endogenic toxins, . . .) microbial contamination (total load, presence of pathogens and spoilage organisms, . . .) contamination with unwanted genetically modified organisms foreign matter (metal or glass fragments, insects, stones, . . .) unwanted matter (nutshells, fruit calices, . . .) Assessing product stability towards • chemical reactions (such as oxidative rancidity) • microbial growth (due to inappropriate pH, water activity, preservative concentration, either in the product as a whole or in a small region within the product) • microbial or chemical contamination (due to defective or inappropriate packaging) (including the migration of compounds in the packaging material into the food) • migration of water or fat (between pastry shell and filling, between food and environment) • loss of protective atmosphere (due to defective seal) (for products packed under a modified atmosphere designed to suppress microbial growth or oxidation)

liquid portion, fat and water may be combined in an emulsion, or even in a double emulsion. Water can be present as free water or bound in many different ways: as water of crystallization, bound to protein or starch molecules, entrapped in biopolymer networks or absorbed on solid surfaces of porous food powder particles. Active enzymes may be present, either in the tissues of fresh meat or produce, or within the cells of the microbial flora. Table 1.3

Measurements in process management

Objectives • ensure safety and continuity of the processing operation • maintain conditions for in-spec. products • use resources efficiently (labour, raw materials, energy, machinery) • reduce loading of effluents (e.g. of waste water with organic matter) Measurements • pressure • temperature (also spatial distribution of temperature and time integral over temperature), • pH • mass and volume flow rates of liquids and particulate solids • fill levels of liquids and particulate solids bulk density, weight apparent viscosity Also wanted, but more difficult to achieve on-line: • chemical composition (gross and fine) • complex rheological properties (yield value, elasticity, . . .) • particle, droplet, bubble size, (average size and distribution) • volatiles evolved in cooking, baking, roasting, drying operations

10

Instrumentation and sensors for the food industry

Table 1.4 • • • •

• •

Instrumentation requirements, on-line

hygienic sensing head contaminant-free (no reagents, no microbes) no foreign body hazard (no fragile glass components), robust CIP (cleaning-in-place) tolerant if permanently installed on-line (alternatives for specific chemical measurements: instruments with disposable sensing element which must be easily replaced and inexpensive, or, in certain applications, robust, easily cleanable dip-probes or stabprobes for in situ measurements) reproducibility in accordance with task, reliable, low maintenance effort suitable for complex chemical and physical sample properties total cost (capital, maintenance, running) in good proportion to benefits

Samples in the food industry are, moreover, very diverse and highly variable. The season, the region of origin, the harvesting and storage conditions as well as the processing steps (such as the fermentation of cocoa beans) will all influence the properties of the raw materials. New food-processing technologies are being introduced to provide an ever wider range of food products that require frequent adaptation to changing consumer preferences and market structures. 1.2.2 Hostile conditions and stringent hygiene requirements The pH extends over a wide range, with low values for vinegar or citrus fruit juices and high values for caustic cleaning solutions used regularly in-line. A wide range of pH values is also encountered in the monitoring of effluents, that is waste liquids formed in washing raw materials or in cleaning container surfaces, for example. Table 1.5

Measurements for special concepts

HACCP – Hazard analysis critical control points On-line measurements • pressure, temperature (spatially resolved, time integral) • relative humidity • product pH product solute content • strength and surface coverage of solutions used for periodic cleaning of machinery Off-line measurements • water activity (as equilibrium relative humidity) • pH (spatially resolved) preservative concentration • microbial contamination of ingredients including water • microbial contamination on machinery and on other surfaces in the production area Marker (indicator) approach For the on-line, at-line or in situ assessment of • microbial pre-spoilage status • oxidative rancidity status • level of heat-induced deterioration • progress of ripening or conditioning browning potential • end of heat processing operation Measured are chemical or physical variables that have first been identified as indicative of the complex condition of interest. Usually, a given marker (or indicator) variable will be valid for a particular group of products only. (See Section 1.3.4 and Chapter 19.)

Instrumentation for food quality assurance

11

Temperatures vary from freeze-drying conditions ( 50ºC or lower) to hot frying fat conditions (up to 250ºC) and roasting operations (320ºC or higher). Processing and packaging under vacuum is employed, and excess pressure is used in cooking and canning operations. A retort would typically operate with pressures of 60–600 kPa, that is 0.6–6 bar (McFarlane 1983, see Appendix B, Tables 3 and 4). Particularly severe conditions can prevail in a cooker extruder, where both high pressures (over 10 MPa, that is, over 100 bar) and high temperatures (around 200ºC) can be encountered. Moreover, the inner barrel surface is scraped by the extruder screw, and access to the food mix within the barrel or in the extruder head is certainly restricted. The food mix itself can be quite abrasive in the early part of its passage through the extrusion cooker. Maize grits, for example, may be present, expanding later on in the fashion of popcorn. These conditions present a challenge even for the design of pressure and temperature (p/T) probes. (For a description of extrusion cookers see McFarlane 1983; O’Connor 1987; Wiedmann and Strecker 1988.) Nevertheless, sensors for the measurement of moisture and other variables are under development for this hostile environment. Radiofrequency open-ended coaxial probes have been designed to fit into the openings foreseen for the bolt-type p/T probes designed for extrusion cookers, and a microwave stripline has been constructed for mounting in an extrusion head (Chapter 9). In general, the conditions in the food industry are more favourable than in a cooker extruder. A common challenge for in situ sensors is, however, the cleaning-in-place (CIP) procedures used in many processing systems in the food industry (Kessler and Weichner 1989). These usually involve flushing with hot caustic soda solutions (NaOH) which can corrode probe surfaces, and this is particularly unfavourable for many chemical sensors. High-pressure steam cleaning is another effective CIP procedure; this will challenge the mechanical and thermal stability of a sensing head. The strict hygiene standards in the food industry also demand that in-line probes in contact with the sample must have crevice-free surfaces. This applies both to the sensing head and to the mounting flange area. For aseptic processes, any sensor surfaces in contact with the sample need to be tolerant to CIP procedures. In fermentation processes, the use of a disposable sterilized sensor can be an option. Any danger of chemical contamination of the food by sensor reagents or components of slight solubility must be eliminated. The introduction of foreign bodies, particularly glass or metal fragments, in the case of damage to the sensor, must also be prevented. Food powders with a very low moisture content can accumulate high electrostatic charges, and sensors that may come into contact with such powders (typically starch-based products) must be designed to minimize the risk of a dust explosion. The transducer and electronics may have to withstand exposure to water, steam or airborne dust. Occasionally, they may be enrobed in chocolate or coated with a thin film of condensed polymerized frying oil. Sensors in contact with food or food volatiles are often subject to fouling by proteins, fats or starch particles (Kessler and Weichner 1989). Electromagnetic interference (EMI) will be encountered in industrial microwave ovens or in direct ohmic heating appliances. There will also be electromagnetic noise and mechanical vibrations from pumps, hoppers and other plant. Rotating mixer paddles scraping the walls of a vessel may be in the way of a radiated signal or restrict the positioning of a wall-mounted probe.

12

Instrumentation and sensors for the food industry

1.2.3 Non-contact techniques and robotic sampling and conditioning Given the often hostile conditions for invasive sensors, either during the processing of foods or during the periodic cleaning operations, and the always stringent hygiene and other food safety requirements, non-contact measurements are particularly attractive to the food industry. These can be based on the interaction of electromagnetic waves, including gamma-rays, light, infrared radiation, microwaves or radiofrequency waves, or of ultrasound signals with the sample. Such methods do, however, require an awareness of the nature of the interaction of the applied signal with the food, its headspace and container. This understanding is needed at all stages of instrument development, in the choice of suitable applications and installation points, during the setting-up and calibration procedures (including the preparation of training samples), in the running of the instrument and in the evaluation of the readings (Chapters 6, 7, 9, 11, 12, 13). To develop the wide range of sensors desirable for in-line measurement in the food industry would be prohibitively expensive. Not only different target variables, but also different analytical ranges and variable chemical and physical environments, would have to be catered for. The recognition of the cost that would be associated with the development of in-line sensors for a wide range of chemical, physical and microbial properties, each for a wide range of diverse applications, has led to an interest in techniques that make the best of the sensors available. Robotic sampling and sample preparation systems allow rapid measurements at short intervals by enabling the use of sensors that would otherwise be confined to laboratory applications. This approach has been implemented particularly in Japan. An example is shown in Fig. 1.3 (see also Chapter 20, Section 5).

1.3

Interpreting the readings

1.3.1 Measured variables and target variables In non-contact measurements, the relationship between the measured variables and the target variables is often complex, so that a given calibration will apply only to a limited range of food products and processing conditions. For instance, a water content measurement based on near infrared reflection analysis will have to rely on a predictable relationship between surface moisture and average bulk moisture content; or, the monitoring of solute concentration by a measurement of ultrasound velocity depends on a

Fig. 1.3 Robotics approach.

Instrumentation for food quality assurance

13

constant composition of both the solute and the carrier liquid (and on compensation for temperature changes, as is the case with most measurement methods). Care in the setting up and calibration are essential, as is the choice of appropriate applications. Non-contact in-line techniques will then provide highly reliable continuous measurements that allow process adjustments before an out-of-specification situation arises (Chapters 6, 9, 12, 13). In contact measurements also, the measured variable is not always the target variable. For example, pH is often measured as an indicator of acid concentration (provided that the acid composition is known). Ion activity is often measured in place of ion concentration. Frying oil samples are taken for an at-line measurement of colour or free fatty acid (FFA) content in order to infer the degree of frying-induced polymerization and oxidation. Yet, both colour and FFA-content are highly dependent on other factors such as the oil type, the food fried and the frying conditions. (See Chapters 19 and 20 on oil quality and pH, respectively.) In the QC laboratory, the assay of chemical composition can involve deductions from the proportion (by weight) of sample becoming volatile or dissolved under certain conditions. Clearly, such assay types, and many others, need to take into account the nature of the sample. Indeed, the official methods prescribe sample-specific assay procedures. In addition to these relationships between measured and target variables, the significance of a target variable for the manufacturing operation needs to be considered. This can reside in assuring the safety and stability of the food, the enjoyment in handling and eating it, and the efficiency in producing it or in complying with legal regulations or customer specifications. These aspects are discussed in Section 1.4.1. 1.3.2 Relationship between in-line and QC laboratory methods Differences between the readings of the in-line instrument and the off-line quality control (QC) laboratory results are, at times, unjustly blamed on the in-line method. The QC reference assay can be applied only to a small fraction of the sample volume passing the sampling point, and this alone can lead to a result differing from an in-line method that provides 100 per cent screening. Moreover, different laboratory methods will often give systematically differing results between them for a particular food sample type, even though they may give identical results for other sample types (see, for example, Fig. 1.4). Often, the definition of the measured variable depends on the laboratory reference method used. For example, for an oven-drying procedure, moisture is defined as that part of the sample that will be driven off at the applied temperature and pressure. For a titration procedure, on the other hand, the relevant part of the water is that which can be extracted from the food matrix (or dissolved and dispersed together with food solids) and brought into contact with the reagent. For this reason, official analysis protocols exist for different food types; for accurate results to be achieved, an instrument used in the QC laboratory needs to be calibrated for each food type against the official method. In some situations, the comparison between the on-line reading and the chemical or rheological QC result, for example, is not strictly valid because the sample changes on removal from the line and during the preparation and analysis steps in the QC laboratory. Oxidation, thinning, thickening, fermentation or other changes can occur during this time, and homogenization steps can cause the breaking up of tissue cells, leading to changes in the composition of the juices and also exposing the cell contents to oxidation or other reactions.

14

Instrumentation and sensors for the food industry

Fig. 1.4 Comparison of laboratory methods for moisture measurement. Oven drying at 102ºC to constant weight leads to an increased apparent moisture content when non-water volatiles are present. This can be avoided by vacuum drying at a lower temperature. When lactose is present, incomplete dehydration can decrease the apparent moisture content by oven drying. This is avoidable by moistening prior to oven drying. Symbols mark different brands of instant coffee and milk powder (Kress-Rogers and Kent 1986).

When changing from a process control system, updated from time to time by data from the QC laboratory, to a system relying primarily on continuous feedback from in-line instrumentation, the significance of the measured properties for the food processing operation needs to be well defined. It may well be favourable to replace an off-line measurement for a given set of variables with an on-line method determining a different set of variables, at a point upstream from the existing QC sampling point. Feedforward control based on the continuous on-line monitoring of the process stream before and during processing can often provide tighter control of the end product properties than delayed feedback control based on an off-line measurement of the end properties themselves. Based on an analysis of the process as a whole, together with an awareness of the options for on-line and at-line instrumentation, an effective data input system for process control purposes can be designed. A complementary range of laboratory instruments will provide calibration updates and a wider range of measurements (see the last entry in Appendix A). 1.3.3 Data processing approaches When several on-line measurements are carried out simultaneously, there is not necessarily a one-to-one relationship between each measured variable (or a combination of these) and a corresponding control parameter to be adjusted. The design of the

Instrumentation for food quality assurance

15

response pattern of the control system can then be based on a variety of data processing approaches. Ideally, but rarely, there is a mathematical model based on sound physicochemical principles, which links the measured variables to the relevant properties of the end product and to the process parameters that can be adjusted. For other cases, data interpretation based on neural networks (Gardner and Hines 1997, Section 5; Jansson 1991) and control decisions based on fuzzy logic (Berrie 1997; Eerikainen et al. 1988) as well as neuro-fuzzy systems (Theisen et al. 1998; Tilli 1996) have become established. (See also the entries on fuzzy logic and neural networks in Appendix A to this book.) Neural networks in particular have attracted much attention in recent years. Here, several on-line sensors that are not specific for an intended property are combined. Extensive output data from these sensors and associated quality control results are then collected over a period of weeks or months. These data are subjected to analysis with neural networks. This can result in a signal processing mode learnt by the neural network. The output of the educated network will then be indicative of a characteristic of the finished product such as taste or aroma, although none of the contributing sensors necessarily has a well-defined physico-chemical relationship to the characteristic of interest. There is a temptation to employ such systems as a replacement for the development of sensors and instruments that give an output with a well-defined relationship to the process and product. However, the short-term saving in avoiding sensor development can be outweighed by the cost of frequent updating of the signal evaluation procedure for nonspecific measurements. This updating procedure may involve the collection of process measurement data and associated product assessments by QC laboratory tests and taste panels over many weeks, and may be required whenever raw materials or recipes change. When the need for an update is not recognized in time, the system will be unreliable, and the quality of the product will be adversely affected. Although sophisticated data evaluation systems will play an important role in the future, they will be complementary to rather than a replacement for measurement techniques based on sound physicochemical principles. (See also Chapter 20, Section 20.5.2.) A special case is the assessment of aroma and taste. Odours and aromas are usually composed of hundreds of compounds carried in an air stream made up of further compounds. Additive, synergistic, antigonistic and compensative effects can occur between the chemical components contributing to a flavour, that is the aroma and taste as perceived by the human nose and tongue. A sensor that is highly selective for just one chemical compound is not meaningful in the assessment of flavour. The standard approach to flavour analysis in the food industry is the organoleptic panel composed of six or more carefully screened and well-trained panellists and following elaborate procedures (see Chapter 2). Due to the cost and time required for this approach, it is suitable for periodic quality control but not for continuous monitoring or field work. To address this problem, electronic noses and tongues have been developed. These are modelled on the mammalian nose and tongue but of lesser complexity, and each of these systems is intended for a specific range of applications. They are based on arrays of (typically about 20) sensors with broadly overlapping specificities combined with pattern recognition methods including neural networks (Chapter 19; Kress-Rogers 1997). 1.3.4 The marker approach; novel sensors In the case of lengthy tests for properties such as freshness, the progress of microbial or oxidative degradation processes, or taste and texture, efforts are increasingly made to

16

Instrumentation and sensors for the food industry

identify and measure chemical or physical indicator variables linked to the condition. This can be described as the marker approach. For example, prototype probes for meat freshness and frying oil quality assessment have been developed, based on a biosensor array and a vibrating rod viscometer, respectively. The assessment of meat and fish freshness by indicator gases and volatiles or by compounds involved in the decomposition of ATP have received increasing attention in recent years (Chapters 19, 20, 22). Novel chemical sensors, including biosensors (Turner et al. 1987; Scheller and Schmid 1992; Kress-Rogers 1997), have been developed in the last four decades, and the pace of this development has gained momentum. A sizeable number of biosensors have already been adapted for food applications. Many other sophisticated biosensor adaptations for the food industry are still at the prototype stage. This is consistent with the delay between the research and application stages of clinical biosensors, for example, which have received attention and funding earlier and more lavishly. In the absence of fully optimized biosensors for food applications, many current practical biosensor applications in the food industry employ the robotic approach (see above). Earlier, the microelectronic pH and ion sensors (primarily ion-selective FETs or ISFETs) had a hesitant commercial development. In recent years, however, commercial instruments designed for the food industry have been available and have solved many of the problems experienced with traditional pH and ion probes (hazard of glass membrane, laborious maintenance, etc.). Fibre-optic probes for chemical sensing are continuing to make progress. They are immune to EMI (electromagnetic interference) in microwave ovens, and can be configured as robust and mechanically flexible remote probes. Optical sensors are also the basis of novel immunosensors (Chapter 20). 1.3.5 Instrumentation as an interdisciplinary subject The development and implementation of successful new measurement applications are tasks requiring effective interdisciplinary communication and cooperation. This can be accomplished only when instrument suppliers and users have a common basis of understanding of the interplay between the physical, chemical and microbial properties of the food and its environment on the one hand, and the physics, chemistry and electronics governing the instrument characteristics on the other. An awareness of the nature of the reference method and of the significance of the target variable for the food processing operation is also needed. In moisture measurement, for example, one has to establish whether the total water content is most relevant, or whether only water molecules with a high degree of mobility play a role. This will depend on whether the measurement is undertaken on account of legal requirements regarding the maximum water content of, e.g. margarine, for the price definition of commodities such as wheat, or with a view to the influence of the water content on the product texture or its relationship with the water activity. The lastmentioned factor determines the stability towards microbial spoilage and water migration from pie filling to pie shell, for example. Water activity is usually determined by a measurement of the equilibrium relative humidity of the sample, which by definition cannot be measured instantaneously. Instead, the water content can be measured in-line, and the water activity can then be deduced for a given product with a given processing history. Moisture and microbes are both ubiquitous, and improved instrumentation for their determination has been a prime concern for the food industry for many years (Chapters 6, 9, 12, 13, 16 on moisture and water activity, Chapters 17, 19–23 on specific microbes and microbial load).

Instrumentation for food quality assurance

17

From the instrument engineer’s point of view, it may seem surprising, at first glance, to find methods based on a physical interaction of the sample with the instrument (using visible, near infrared, microwave or ultrasound waves, or employing mechanical stress; Chapters 3–9, 12–15, 18) and those based on chemical interactions (Chapters 19–23) presented together. The food scientist, on the other hand, may prefer to consider instruments for the determination of physical properties (rheology, colour, density, particle size) of the sample separately from those for the assay of chemical or microbiological properties. Process control engineers may wish to separate on-line and at-line instrumentation from off-line instruments. However, in formulating a system of quality assurance and quality control procedures for a food manufacturing process, it is advantageous to be aware of the alternative techniques, each with its own characteristics, for the determination of a particular sample property (for example, for sugars, see Chapters 6, 8, 12, 13, 20–23; for water content see Chapters 6, 9, 12, 13). A number of instrument types allow the measurement of both physical and chemical properties. For example, bulk density and water content can be determined simultaneously by a microwave technique (Chapter 9; Kress-Rogers and Kent 1987); film thickness and composition can be measured with near infrared instruments (Chapter 6). Also, there is often a choice between process control based on measuring a certain set of variables upstream and in-line (for example, pressure, temperature, pH), and that based on measuring another set downstream and off-line (for example, glucose/sucrose ratio). As a further option, some off-line instruments can be used at-line with the help of robotic sampling and sample preparation, thus providing rapid feedback based on a wider range of measured variables. On-line and QC laboratory methods have to interlink in guiding process control, and are therefore presented together in this book, with the emphasis on the interplay between the instrumental method, the food and the manufacturing process.

1.4

Measurement types

1.4.1 Target variables The on-line or rapid at-line measurement of variables relevant to the eating quality, the wholesomeness and the safety of the food product is an increasing concern of the food industry, and this is reflected in the topics covered in this book. Colour and other aspects of the appearance are dominant in determining the first impression and influence the choice of food products by the consumer. Aroma, taste, and texture or mouthfeel influence the enjoyment of the food and determine whether the consumer comes back for more of the product. The rheological properties also affect the handling characteristics of the product both in processing and in the hands of the consumer. All these food properties can be assessed by panels of trained persons, or animals in the case of petfood (Chapter 2). However, for frequent control purposes this is not practical, and instruments for the measurement of colour attributes, of variables related to the chemical composition (either proximate analysis or more specific and sensitive measurements) and of those related to the physical structure of the food (rheology, particle size, new methods in texture assessment) are used if equipment of the desired specifications is available (Chapters 3–5, 6, 8, 14, 15, 18, 20–23). Electronic noses and tongues have been developed in recent years to mimic the human (or pet) sense of smell and taste for specific applications (Chapter 19). The assessment of wholesomeness and safety is not generally possible or advisable with the human senses, and instruments or laboratory tests will always be required for

Colour Sorting by colour Temperature Temperature – time integral Pressure Level and flow rate Particle, droplet or bubble size Solid/liquid ratio and crystal size Bulk density Rheology Texture Water activity Water content Proximates: fat, protein, carbohydrate, ash pH Acidity Sodium, potassium, calcium Specific sugars 

 5





 5



6

  

Chemical aspects

  

Microbial aspects









 

 

 



4  





 5



 

 

Appearance



Nutritional aspects





  

Physical aspects

Food safety and stability

Table 1.6 The significance of measurements for the manufacturing operation



   4  









Texture consistency

Food quality

i 

i

4





Aroma



i  

4  





Taste

 

 3

 2

Ingredients management



     



  



Process management

Production efficiency



  





Legal conditions



  

     







 

Customer specifications

Compliance



1

1  



1



Labelling regulations

    



 









6





6

  







  8

  







8

7 









  









8

7









7   8





  





 

 









 

  













  

 1 1









 

Authenticity assessment. The comparison of the sample’s free fatty acid composition, ultraviolet absorption spectrum or other ‘fingerprint’-type assays with the pattern expected for the claimed food type and origin is used in the authenticity assessment of foods.

Key  direct link; i strong indirect link; 1 influencing storage conditions or use-by date; 2 influencing packing density; 3 influencing enrobing thickness; 4 governing microbial growth and metabolism as well as chemical reactions and water migration processes, thereby influencing composition and texture; 5 linked to water activity for a given food with a given process history; water content is more easily measured at-line and can be used as an indicator of water activity under controlled conditions; 6 solutes such as sugars or salts reduce water activity and have traditionally been used to preserve foods (for example jam, bacon); 7 at high temperatures, reducing sugars react with amino acids to form a wide range of compounds that flavour and colour roasted and baked products (Maillard reaction); 8 for example, antioxidants reduce the rancidity of oils and the browning of produce.

Total reducing sugars Solute concentration Alcohols Preservatives, antioxidants Vitamins, trace minerals Emulsifiers, gelling agents Flavourings, colourings Toxins, residues Volatiles during cooking Volatiles during storage Microbial load, contamination Biomass, functional Chilled meat freshness Frying oil quality

20

Instrumentation and sensors for the food industry

these food properties, including the content of nutrients, microbial load, chemical contaminants, pathogenic microbes and foreign bodies. Not only the current status but also the stability of the product during the intended shelf-life needs to be assessed. (On the shelf-life of foods, see for example Chapter 7 in Stewart and Amerine 1982.) To this end, a determination of the product’s water activity, pH, packaging integrity, existing flora, content of preservatives, antioxidants content and other factors is undertaken. The temperature of a food needs to be monitored in heat processing (such as sterilization), tempering and conditioning and also in storage. Spatial resolution is often needed, particularly information on the hottest and coldest points of the sample. A timetemperature profile including integral, maximum and minimum is also an important measurement target. Remote infrared thermometry has the advantage of providing a noncontact measurement and is also the basis of thermal imaging systems (Chapter 7). A development for depth profiling is the distributed fibre-optic temperature probe. Table 1.6 shows the significance for the manufacturing operation of measurements of a range of variables. 1.4.2 Instrumental methods On-line instruments based on the interaction of electromagnetic waves or ultrasound with the sample are discussed in the first part of the book. In the second part, instruments in the quality control laboratory are treated. In the third part, the expanding range of chemical sensors, including biosensors, is presented. The development of sensors for the identification of an indicator variable that is more amenable to on-line implementation than current chemical reference methods for quality assessment is discussed. The assessment of sensory properties by both sensory panels (composed of humans or pets) and by instruments (optical instruments, electronic noses and tongues, texture analysers) are detailed. Instruments that require an understanding of their interaction with the food and the manufacturing process as a condition for successful application are the subject of this book. In order to provide a useful description of principles and food-industry-specific application details for such instruments, it has been necessary to choose certain instrument types to the exclusion of others. Only instrumental methods that can be understood fully on the basis of classical physics have been included in this book; thus instruments based on nuclear magnetic resonance (NMR) or electron spin resonance (ESR) are not represented. This is not to imply the lesser importance of these methods. In fact, the development of new applications of NMR and NMR imaging (NMRI) to the food industry has progressed rapidly, and provides the basis for many new instruments (see, for example, Levine and Slade 1991, pp. 405–626). For the recognition of irradiated foods, ESR applications are being developed. Pulsed NMR techniques are already widely used in quality control in the food industry for the determination of water and lipids content and of the solid/liquid ratio of fat. Bench-top instruments for these applications are readily available and are being further optimized for a wider range of food applications (including the investigation of water ‘binding’ to biopolymers) and more convenient operation. Furthermore, research into the NMR characteristics of foods and the development of novel magnet designs for NMR instruments are now underway. This will lead to new application areas for NMR measurement in the food industry. One area is the on-line implementation of bulk NMR measurements as they are currently undertaken in QC laboratories. These are to become feasible for a much wider range of process lines than is possible at present. They will aid

Instrumentation for food quality assurance

21

in the control of many operations in food processing, including baking, drying, concentrating, freezing, thawing and tempering. The other area is the use of NMR imaging in product development and quality control tasks. This latter measurement type will, in the medium term, be restricted to central laboratories owing to the cost and complexity of the equipment. Also omitted in this book are those instruments that rely on ionizing radiation (such as x-ray foreign body detection systems) or on a substantial computing capacity, as, for example, image acquisition and analysis systems. Real-time processing and automatic evaluation of visible light, x-ray, infrared or other images is another area that is expanding rapidly. The underlying principles and application details relevant to the food industry for these systems could best be presented in a book dedicated to this area alone. The appendix to this chapter summarizes measurement types. 1.4.3 Fringe benefits The dielectric properties of food, discussed here in the context of measurements based on microwaves, are also relevant in the application of such radiation at much higher intensities to the processing of foods in industrial and domestic microwave ovens (Ohlsson 1988). Indeed, many dielectric data for foods have been acquired in order to predict microwave heating characteristics. For food scientists, a familiarity with the principles underlying dielectric measurement affords the added benefit of helping in the analysis of processing characteristics. To a lesser extent this applies to ultrasound, which also has applications both as a low-intensity signal used in measurement and as a highintensity irradiation employed in sonoprocessing and sonochemistry. The overlap between the food properties relevant for measurement and those relevant for processing is smaller here. Ultrasound processing is therefore briefly discussed as a separate topic (Section 13.2.7).

1.5

Further reading

The emphasis in this book is primarily on the measurement of food properties as this requires a particular understanding of the interplay between the instrumental method and the food and its environment. However, this understanding is also required in the measurement of certain process variables such as the volume and mass flow rates of food liquids with complex rheologies. Ultrasonic measurement of volume flow rate and of liquid and solid level is discussed in this book (Chapters 11–13). Volume flow rates do not always provide adequate information for materials with uneven aeration or temperature distribution, and are also difficult to measure for nonNewtonian liquids (Chapters 14, 15) or inhomogeneous materials. For materials such as molten chocolate, mass flow rate can now be measured successfully with Coriolis force flow rate meters (developed by Exac and Micromotion). These have the form of a Ushaped tube (or two parallel U-tubes), which can be flanged into the pipeline without restricting the flow. The Coriolis technique has been reviewed by McKenzie (1990), with details on one of the two embodiments. A straight vibrating tube section can be flanged into pipes for measuring the mass density of the process stream. The tube section is isolated vibrationally from the remainder of the pipe by bellows. Another sensor based on mechanical resonance is the vibrating rod probe for the measurement of level, density or viscosity. Such probes are

22

Instrumentation and sensors for the food industry

available in dipstick form or in flange mountings. This sensor family is represented in the book in the form of a prototype probe for the in situ determination of frying oil quality, based on a measurement of the viscosity of the hot oil by a vibrating tube viscometer (Chapter 19). A review of the principles underlying the function of a wide range of mechanical resonance sensors has been given by Langdon (1985). For the measurement of temperature, both contact methods and a non-contact method that also lends itself to combination with linescan and imaging techniques is presented in this book (Chapter 7, 10). Fibre-optic thermometers and quartz temperature sensors (shear mode resonators or tuning fork devices), which have become available for industrial applications in recent years, are reviewed by Schaefer (1989). Updated overview tables on the specifications of conventional contact thermometers such as thermocouples are published regularly by transducer magazines. This is also the case for strain gauge pressure transducers. Information on special miniature designs for pressure and temperature transducers that can be mounted inside sample cans during heat processing can be obtained from the suppliers. Humidity sensors have been discussed as parts of instruments for the measurement of water activity. Similar sensors can also be used in monitoring ambient humidity in many storage or processing areas. In the latter applications, the required accuracy is less stringent; however, the emphasis is on robustness and long-term stability with minimal maintenance. High temperatures often need to be tolerated, for example in oven flues (McFarlane 1983). Inexpensive miniature humidity sensors based on solid-state devices are now part of a number of domestic appliances (such as some makes of tumble dryers or microwave ovens). An overview of the automatic control of food manufacturing processes is given by McFarlane (1983). He describes processes in the areas of raw materials handling, recipe dispensing, pre-processing, cooking processes, biochemical processes, finishing and packaging. The range and tolerance for the controlled variables are given for each process. There are also books giving details on processing operations in one particular sector such as fish canning (Wheaton and Lawson 1985). Some recent developments in the control and optimization of food processes are presented in symposium proceedings edited by Renard and Bimbenet (1988) and by Spiess and Schubert (1990). A comprehensive compilation of instruments, each described briefly without industryspecific aspects or application details, is available in the form of a handbook on general instrumentation (Noltingk 1988). Often, instrument manufacturers will supply application details and related literature. An introduction to food science and food technology has been given by Stewart and Amerine (1982). Information on the physical properties of foods can be found in the books by Lewis (1987), Jowitt et al. (1987), Singh and Medina (1988) and Okos (1986).

1.6

References

and HILLIAM, M. (1991) The UK food and drinks report: market, industry and new product trends. Leatherhead Food Research Association Special Report, April 1991. BERRIE, P.G. (1997) Fuzzy Logic in the Evaluation of Sensor Data, pp. 469–500 in Handbook of Biosensors and Electronic Noses: Medicine, Food and the Environment, Kress-Rogers E. (ed.), Boca Raton, New York, London, Tokyo, CRC Press Inc. BAILEY, L., BOYLE, C.

Instrumentation for food quality assurance

23

and ATKINSON, T.A. (1990) Oscillometric instrument for the non-invasive detection of low-level microbial activity. Biosensors and Bioelectronics, 5, 273–89. EERIKAINEN, T., LINKO, S. and LINKO, P. (1988) The potential of fuzzy logic in optimization and control: fuzzy reasoning in extrusion cooker control, pp. 183–200. In Automatic Control and Optimization of Food Processes, Renard, M. and Bimbenet, J.J. (eds) Elsevier Applied Science, London. GARDNER, J.W. and HINES, E.L. (1997) Pattern Analysis Techniques, pp. 633–52 in Handbook of Biosensors and Electronic Noses: Medicine, Food and the Environment, Kress-Rogers E. (ed.), Boca Raton, New York, London, Tokyo, CRC Press Inc. GEORGALA, D.L. (1989) Modern food processing, in Food Processing, Proceedings of the Ninth British Nutrition Foundation Branch Conference, Cottrell, R.C. (ed.) Parthenon Publishing Group, Carnforth and New Jersey. JANSSON, P.A. (1991) Neural networks: an overview. Analytical Chemistry, 63, 357A– 362A. JOWITT, R., ESCHER, F., KENT, M., MCKENNA, B. and ROQUES, M. (eds) (1987) Physical Properties of Foods Vol. 2, COST 90bis Final Seminar Proceedings. Elsevier Applied Science, London and New York (arranged by the Commission of the European Communities). KESSLER, H.G. and WEICHNER, K. (eds) (1989) Fouling and Cleaning in Food Processing, Proceedings of the Third International Conference on Fouling and Cleaning in Food Processing. Druckerei Walch, Augsburg. Distributed by: Heinz G. Kessler, Institut fu¨r Wissenschaft und Verfahrenstechnik in der Milchverarbeitung, Technische Universita¨t Mu¨nchen, D-8050 Freising-Weihenstephan, Germany; Daryl B. Lund, Department of Food Science, PO Box 231, Rutgers University, New Brunswick, NJ 08903, USA. KRESS-ROGERS, E. (1985) Technology transfer. II: The new generation of sensors. Leatherhead Food Research Association Scientific and Technical Survey, 150. KRESS-ROGERS, E. (1986) Instrumentation in the food industry. Part I: Chemical, biochemical and immunochemical determinands. Part II: Physical determinands in quality and process control. Journal of Physics E: Scientific Instruments, 19, 13–21, 105–9. KRESS-ROGERS, E. (ed.) (1997) Handbook of Biosensors and Electronic Noses: Medicine, Food and the Environment, Boca Raton, New York, London, Tokyo, CRC Press Inc. KRESS-ROGERS, E. and KENT, M. (1986) Two-parameter microwave technique for measurement of powder moisture and density. Leatherhead Food Research Association Research Report, 553. KRESS-ROGERS, E. and KENT, M. (1987) Microwave measurement of powder moisture and density. Journal of Food Engineering, 6, 345–76. LANGDON, R.M. (1985) Resonator sensors – a review. Journal of Physics E: Scientific Instruments, 18, 103–15. LEVINE, H. and SLADE, L. (1991) Water Relationships in Food, Plenum Press, New York and London. LEWIS, M.J. (1987) Physical Properties of Foods and Food Processing Systems. Ellis Horwood, Chichester and VCH, Weinheim. MCFARLANE, I. (1983) Automatic Control of Food Manufacturing Processes. Applied Science Publishers, London and New York. MCKENZIE, G. (1990) Mass flow measurement. Sensor Review, July, 129–32. COSSAR, J.D., BLAKE-COLEMEN, B.C., RAMSAY, C.

24

Instrumentation and sensors for the food industry

NOLTINGK, B.E. (ed)

(1988) Instrumentation Reference Book. Butterworth-Heinemann,

Oxford. (1987) Extrusion Technology for the Food Industry. Elsevier Applied Science, London and New York. OHLSSON, T. (1988) Dielectric properties and microwave processing, pp. 73–92 in Food Properties and Computer-Aided Engineering of Food Processing Systems, Proceedings NATO Workshop, Porto, Singh, R.P. and Medina, A.G. (eds), Kluwer, Dordrecht, Boston and London. OKOS, M.R. (ed.) (1986) Physical and Chemical Properties of Food. American Society of Agricultural Engineering, Michigan. RENARD, M. and BIMBENET, J.J. (eds) (1988) Automatic Control and Optimization of Food Processes. Elsevier Applied Science, London and New York. SCHAEFER, W. (1989) Temperature sensors: new technologies on their way to industrial application. Sensors and Actuators, 17, 27–37. SCHELLER, F. and SCHMIDT, R.D. (1992) GBF Monograph Biosensors: Fundamentals, Technologies and Applications, Marcel Dekker, New York. SINGH, R.P. and MEDINA, A.G. (1988) Food Properties and Computer-Aided Engineering of Food Processing Systems, Proceedings NATO Workshop, Porto, October 1988, Kluwer, Dordrecht, Boston and London. SPIESS, W.E.L. and SCHUBERT, H. (1990) Engineering and Food, vols 1–3, Elsevier Applied Science, London. STEWART, G.F. and AMERINE, M.A. (1982) Introduction to Food Science and Technology. Academic Press, New York. THEISEN, M., STEUDEL, A., RYCHETSKY, M. and GLESNER, M. (1998) Fuzzy Logic and NeuroSystems Assisted Intelligent Sensors, pp. 29–59 in Sensors Update Volume 3, Baltes H., Go¨pel W. and Hesse J. (eds), Wiley-VCH. TILLI, T. (1996) Building Intelligent Systems with Fuzzy Logic and Neural Networks, John Wiley and Sons, New York. TURNER, A.P.F., KARUBE, I. and WILSON, G.S. (1987) Biosensors: Fundamentals and Applications, Oxford University Press, New York. VIDAL, P. (1988) Automatization and optimization des proce´de`s de l’industrie alimentaire, pp. 3–16 in Automatic Control and Optimization of Food Processes, Renard, M. and Bimbenet, J.J. (eds), Elsevier Applied Science, London. WHEATON, F.W. and LAWSON, T.B. (1985) Processing Aquatic Food Products. Wiley, New York. WIEDMANN, W. and STRECKER, J. (1988) Process control of cooker-extruders, pp. 201–14 in Automatic Control and Optimization of Food Processes, Renard, M. and Bimbenet, J.J., (eds), Elsevier Applied Science, London. O’CONNOR, C.

Instrumentation for food quality assurance

25

Appendix: Measurement types Measurement types discussed in this book Interaction of electromagnetic waves with foods or containers Range description

Order of magnitude of: frequencies (equivalent energy) wavelengths

Measurement applications

Soft ultraviolet, visible, near infrared

 1015 to 1014 Hz  100 nm to 1 m

(Authenticity of foods) Sorting by ‘colour’

Visible

 1015 Hz 400 to 700 nm

Colour (Optical imaging, for example to identify defective items or to measure dimensions)

Near infrared

1014 Hz (equivalent temperature of radiating black body:  1000 K) 700 to 2500 nm

Water content Fat, protein and carbohydrate content Caffeine contents of 1% m/m or higher Thickness of coated or laminated films on packaging materials

Mid infrared

1013 Hz (equivalent temperature of radiating black body:  100 K) 2.5 to 30 m

Volatiles in headspace Authenticity Identification of ingredients

Near and mid infrared

1 to 15 m

Remote temperature measurement Thermal linescan (Thermal imaging)

Microwaves

 1011 to 109 Hz  1 to 10 cm

Water content of powdered or granular material Water content of low- and intermediate-moisture foods Water content of highmoisture foods Simultaneous bulk density measurement Fat content Ratio polar/non-polar liquids content

26

Instrumentation and sensors for the food industry

Radiowaves, upper range

 108 to 107 Hz  1 to 10 m

Water content Salt content Particle size, shape, distribution Bulk density

Note: Measurements based on ionizing radiation are not included here (for these see later in this appendix). Some applications involving spectroscopic or imaging techniques (not covered in this book) are given in parentheses in the column ‘Measurement applications’ in the table above. Interaction of ultrasound waves with food or container Common frequency and wavelength, order and magnitude Path through gas:  100 kHz  3 mm Path through liquid:  1 MHz  1:5 mm Path through solid:  10 MHz  0:6 mm Applications: Fill level of liquids or solids Volume flow rate of liquids, gases and steam Flow obstructions Location of interfaces between: liquid and second liquid; or liquid and foam; or fat and lean meat (including the back fat thickness of live animals) Solute content of a liquid Discrimination between two liquids (sensor can even be at a stand-off distance from the closed container) Scan of pipe cross-section for liquids carrying solid items Characterization of sheet-type solid passing on a moving belt Counting of food packs Suspended solids (in liquid) mass fraction Dispersed droplets or bubbles (in liquid) volume fraction Size distribution of dispersed solids, droplets or bubbles (in liquid) Creaming, sedimentation Melting, crystallization Solid/liquid ratio of fats Temperature Density of a liquid (as Z/c: see Chapter 13) Tissue structure Acoustic emission monitoring Range: Audiosound and ultrasound Applications: Powder flow Drying Grinding Foam evolution Cleaning fluid impact Cooker extruder performance

Instrumentation for food quality assurance

27

Interaction with chemical sensors, including biosensors and immunosensors Principle: Ion-sensitive membrane, inorganic catalyst, enzyme, antibody, specific adsorbent, etc. immobilized on base transducer, with electromechanical, optical, thermal or acoustic base. Applications: pH (by ISFET) Inorganic ions (by ISFET) pH (by fibre-optic probe) Acidity (by microtitrator, ISFET-based or by electronic tongue) Gases (by CHEMFET, MeOx or organic semiconductors) Specific sugars, starch, alcohols, amines, organic acids, amino acids, essential fatty acids, lipids, etc. (by biosensors) Vitamins, toxins (bacterial, fungal, algal) and specific microbes (by immunosensors) Residues (pesticides, antibiotics, hormones) by immunosensors Specific microbes (by DNA probes) Genetically modified organisms, transgenic crops (by DNA probes, by immunosensors for novel proteins) Meat species (by immunosensors) Fish freshness (compounds involved in ATP decompostion by biosensors) Under development: biosensors for extreme values of pH, pressure, temperature (based on molecularly imprinted polymers) Electronic noses and tongues, marker approach Aroma (volatiles to which the human or pet nose responds) Taste (sour, salty, bitter, sweet, umami) Titratable acidity Classification by brand, quality, etc. Authenticity Ripeness Freshness Conditioning Fermentation monitoring Monitoring of roasting Sensory properties by human/pet sensory panels and by instrumental methods Method: Determined property: Panel or transmittance, reflectance, Appearance (colour, gloss, shape) particle sizing Flavour (aroma, taste) Panel or electronic noses and tongues Chemical analysis (GC-MS, titration, . . .) Panel or texture analysers or rheology Texture, mouthfeel, pouring characteristics Acoustic emission monitoring Sounds during preparation and consumption (sizzling, crackling, crunching)

28

Instrumentation and sensors for the food industry

Monitoring headspace humidity or broth impedance Method: Determined property: Equilibrium relative humidity (ERH) Water activity (stability towards microbial measurement growth and chemical reactions) Electrical impedance of culture broth (at Microbial load 2–10 kHz) monitoring (as method in rapid microbiology) Properties under mechanical stress Method: Rheological analysis Texture assessment Mechanical resonance dipstick probes Resonating tube section Coriolis force meter Mechanical flow rate meters: Positive displacement flow meter Differential pressure method Turbine flow meter Vortex shedding method Manometric method

Determined property: Viscosity, yield point Changes of the rheological properties with shear rate and time Texture attributes (crisp, juicy, rubbery, mealy, tough, etc.) Level, density, viscosity, flow rate Determination of frying oil quality Mass density of a processing stream Mass flow rate of liquid or gas Liquid volume flow rate Liquid, gas or steam flow Pressure

Monitoring process parameters with AC/DC measurements Sensor/instrument/method: Determined property: Pressure Piezoresistor Strain gauge on diaphragm Capacitive pressure sensor on diaphragm Temperature of the process stream (whether Bimetallic strip the contact measurement is representative RTD (resistance temperature detector) Themocouple probe for the bulk of the food will depend on the thermal conductivity and heating pattern) Strain gauge configured for hydrostatic Level of liquids head measurement Electrical capacitance Liquid level Conductance measurement Electrical capacitance measurement Solids level Electromagnetic flow meter Flow rate of conductive liquids Electrical conductivity measurement CIP cleansing agent concentration Electrical conductivity measurement Water quality (simple electrode pair at 1–5 kHz)

Instrumentation for food quality assurance

29

Further measurement types (not covered in this book) Observation of drying, extraction, reaction and chromatographic separation Method: Analyte: Drying or titration Water content Combustion Ash content Solvent extraction Fat content Acid digestion, distillation, and then Nitrogen ! protein content titration Combustion, and then thermal Nitrogen ! protein content conductivity detection HPLC (High Performance Liquid Specific sugars Chromatography) Ions (nitrite, nitrate, bromide, chloride, sulphate) IC (Ion Chromatography) Sulphite GLC (Gas-Liquid Chromatography) Pesticides GC-MS (Gas Chromatography – Mass Taints, flavours, spoilage indicators, toxins Spectrometry) Interaction of electromagnetic waves with foods or containers Applications using ionizing radiation are listed here. Range description

Order of magnitude of: frequencies (equivalent energy) wavelengths

Measurement applications

Gamma-rays (hard Roentgen rays

 1020 to 1019 Hz (equivalent energy:  1 meV to 100 keV)  1 to 10 pm  1018 to 1017 Hz  100 pm to 1 nm ˚) (100 pm ˆ 1 A

Density

X-rays (soft Roentgen rays)

Foreign body detection (with linescan or imaging techniques)

Further measurements probing the properties of foods Method: Determined property: Nuclear magnetic resonance (NMR) Water or fat content Solid/liquid ratio Electromagnetic induction Metallic foreign body Electrical conductivity (multi-electrode Solution strength system or inductive coupling to overcome deposits) Note: The measurement of water content and solution strength is discussed in Chapters 6, 9, 12, 13; that of solid/liquid ratio in Chapter 13. The separation of foreign bodies by optical sorting is described in Chapter 5; impedance monitoring as a rapid microbial method in Chapter 17.

30

Instrumentation and sensors for the food industry

Instruments not interacting with complex food properties Sensor: Determined property: Humidity sensors, based on a Ambient humidity measurement of: polymer or oxide capacitance (not suitable in ovens where ammonia and volatile oils are present) dew point Fibre-optic probe (with fluorescent Temperature compound on tip) Load cell arrangements Container weight Note: Remote thermometry is covered in Chapter 7, ultrasonic flow rate and level metering in Chapters 12, 13, the measurement of equilibrium relative humidity in Chapter 16.

2 Instrumental measurements and sensory parameters A. Hugi and E. Voirol, Nestle´ Research Centre, Lausanne

2.1

Introduction

The human senses have always been used to assess food quality. For centuries they were the only instruments available, until scientific advances in physics, chemistry and biology, as well as the growing demands of the food processing industry, led to the development of analytical techniques aiming at understanding and controlling all aspects of food quality. Although the senses of sight, hearing, taste, smell and touch are used daily in all aspects of our lives, their analytical application to evaluate food properties is relatively recent. Since this discipline is based on the perceptions and judgements of persons, with their inherent inter-individual and temporal variability, it is not surprising that the earliest contributions came from the fields of psychology and statistics. The nineteenth century saw the development of a scientific domain called psychophysics, concerned with the understanding of how people react to external stimuli. The focus was therefore less on the object being evaluated than on the subject evaluating it. Nevertheless, several current sensory test methods stem from that period (Peryam, 1990). Modern sensory evaluation, defined as ‘. . . a scientific discipline used to evoke, measure, analyze and interpret reactions to those characteristics of foods and materials as they are perceived by the senses of sight, smell, taste, touch and hearing.’ (Anon., 1975), really started to grow between the 1930s and the 1950s, with significant advances originating from governmental organisations, scientific or trade associations, and private companies (Peryam, 1990). Since then, sensory scientists have continued to develop new methods and to refine existing ones, but the food industry as a whole has been relatively slow to recognise the usefulness and validity of modern sensory methods. Over the last decades however, increasing competitive pressure has led food companies to pay more attention to consumer needs and preferences. Today, foods from major suppliers usually offer comparable safety and nutritional value. Thus sensory characteristics are increasingly important for consumers as the differentiating factor between foods and brands. As a result, food companies recognise more and more the necessity to measure, optimise and control the sensory properties of

32

Instrumentation and sensors for the food industry

Table 2.1

Examples of practical questions answered by sensory tests

Is this reduction in viscosity perceivable?

Difference test

What is the effect of modifying the recipe (or the process) on a product’s sensory properties?

Descriptive test

Can we substitute a specific ingredient by another one without affecting the sensory characteristics?

Difference test

What are the differences in sensory terms between our product and its competitors?

Descriptive test

What is the ideal sensory profile for this group of consumers?

Descriptive test combined with consumer test (preference mapping)

Is this product batch within specification?

Quality control test

How long can the product be stored before a significant change in sensory quality is noticeable?

Difference (or descriptive) test

their products. At the same time, the competitive pressure forces companies to master their cost structure, and to be more productive with fewer resources. Sensory evaluation is a good way to save money safely, for instance by shortening the development time of a new product or by reducing the amount of rework in a factory. This pressure on costs implies also a permanent search for efficient sensory methods delivering fast, businessoriented results. Sensory evaluation offers today an array of analytical techniques with wide-ranging applications in research, market surveys, product development, and quality assurance. It is used to answer many practical questions arising in a food company, such as the examples shown in Table 2.1. Given the high demand in time and resources of sensory evaluation, it is of interest to identify instrumental methods to replace it, particularly for routine quality control in a factory environment. Williams (1994) listed a vast array of chemical and physical information that can be related to sensory properties (Table 2.2). What is important to keep in mind is that a direct, causal relation between a physical/chemical measurement and a perceived sensory property cannot be simply assumed, it needs to be assessed and validated by a careful experimental study, carried on within the food product range of interest. While the concentration of sugar in a water solution has a univariate relationship with sweetness perception, this is probably no longer the case as the stimulus system becomes more complex, as in a confectionery bar, for example.

2.2

The nature of sensory perception

2.2.1 Senses as measuring instruments Information from the outside world is received via our senses and processed by our brain to determine our reaction and behaviour in response to these signals. The sense organs, together with their related operational physiology, and neural signal processing systems, are evolved structures which were moulded by natural selection operating on environmental selective pressures (Stoddart, 1999). We are sensitive to two kinds of stimuli, physical (heat, pressure, acoustic waves, and electromagnetic radiation) and

Instrumental measurements and sensory parameters Table 2.2

33

Chemical/physical measurements relating to sensory properties (from Williams, 1994)

Measurement Light transmittance/reflectance • visual spectra • tristimulus values • ultraviolet/near infra-red Size and shape • grading • visual analysis Gas chromatography • volatile components Liquid chromatography • non-volatile components Simple chemical/physical measurements • pH, specific gravity, titratable acidity • density, fibre content • alcohol-insoluble solids • moisture • sulphur dioxide Mechanical deformation • shear measurement • compression measurement • puncture measurement Mastication Ultrasonics Molecular characteristics • bond length • spectroscopic information • stereochemistry

Related property Colour Chemical composition, general quality Appearance General quality Aroma, flavour Colour, taste, flavour Taste, general quality Texture General quality Texture General quality Texture Texture General quality Molecular aspects of aroma and taste

chemical. The senses of sight, hearing and touch react to physical stimuli, while taste, smell and trigeminal receptors react to chemical stimuli. 2.2.2 Physical senses Vision The first contact we have with a food product is often visual. Colour and appearance guide us in the purchase of fresh products like fruits or vegetables, and food manufacturers pay due attention to the visual elements of packaging in order to appeal to consumers. In food processing and cooking, colour serves as a cue for the ‘doneness’ of foods and is correlated with changes in aroma and flavour (Lawless and Heymann, 1998). The reader is referred to Chapter 3, by D. B. MacDougall, for detailed information on the physiological basis of colour vision and on the means to measure and communicate colour information. Beside colour, the assessment of food appearance includes many other visual characteristics that are related to physical properties: gloss, transparency, haziness, turbidity, uniformity of colour or surface, size and shape. Hearing This sense is of minor importance for the evaluation of foods, but it is not negligible: the noises of a cork popping out of a wine bottle or of a steak sizzling in the pan are signals

34

Instrumentation and sensors for the food industry

that we easily recognise and react to, even in absence of other sensory cues. Common noise characteristics of foods are crispiness, crunchiness and squeakiness (Meilgaard et al., 1999). In sensory terms, sounds are characterised by their pitch (frequency, Hz), intensity (loudness, dB) and persistence. Physical vibrations in the air, in the approximate range of 16 Hz to 20 kHz, cause the eardrum to vibrate. The vibrations are transmitted via three distinctively shaped bones in the middle ear to create hydraulic motion in the fluid of the cochlea, in the inner ear. The cochlea is a spiral canal covered in hair cells of various lengths which, when agitated, send neural impulses to the brain. When we masticate foods, a large proportion of the sound travels by bone conduction through the teeth, jaws and bones directly to the cochlea (Vickers and Bourne, 1976). Touch The group of perceptions generally described as the sense of touch can be divided into ‘somesthesis’ (tactile sense, skinfeel) and ‘kinaesthesis’ (deep pressure sense or proprioception), both of which sense variations in physical pressure (Meilgaard et al., 1999). Various types of nerve endings in the skin are responsible for the somesthetic sensations we call touch, pressure, heat, cold, itching, and tickling. The particular sensitivity of the lips, tongue and hands enables us to perceive minute differences in the geometrical (gritty, chalky, fibrous, e.g.) and moisture (juicy, oily, greasy, e.g.) properties of foods. Kinaesthesis is felt through nerve fibres buried in muscles, tendons and joints. We assess the mechanical properties of foods – hardness, cohesiveness, adhesiveness, denseness, springiness – via kinaesthetic perception of the stress exerted by muscles of the hand, jaw, or tongue, and of the resulting strain (compression, shear, rupture) within the food sample being handled or masticated. 2.2.3 Chemical senses Olfaction Olfactory receptors are located in a small patch of specialised epithelium in the roof of the nasal cavity. In this so-called olfactory epithelium, millions of neurones carry on one end several hair-like cilia immersed in mucus. Odour molecules bind to specialised receptor proteins within the ciliary membrane. It has been determined that mammals have around 1,000 different odour receptors, although they are capable of detecting at least 10,000 odours. Consequently, each different receptor must respond to several odour molecules and each odour must bind to several receptors. At the other end of each neurone, a fibre known as an axon runs into the olfactory bulb of the brain. The neurones expressing a given receptor are randomly distributed throughout the epithelium, but project their axons to one or, at most, a few sites called glomeruli, in the olfactory bulb. As the positions of the glomeruli are topologically defined, the olfactory bulb provides a two-dimensional map that identifies which of the numerous receptors have been activated in the nose (Axel, 1995). The dynamic range of olfactory receptors, between the threshold concentration and that producing saturation, is small compared to other senses like hearing and sight. On the other hand, the nose has enormous discriminating power: a trained perfumer can identify 150 to 200 different odour qualities. Moreover, olfactory receptors can be 10- to 100-fold more sensitive to some chemicals (e.g., allyl mercaptan: 6  107 molecules per ml of air) than the most sensitive gas chromatograph (ca. 109 molecules per ml) (Meilgaard et al., 1999).

Instrumental measurements and sensory parameters

35

Gustation Taste receptor cells are clustered into taste buds, onion-shaped structures embedded within the lingual epithelium in the fungiform, foliate and circumvallate papillae, and found also in the soft palate and the epiglottis (Dulac, 2000). An adult has about 2,000 taste buds, about half of which are situated on the circumvallate papillae (Plattig, 1988). Each taste bud contains 30–50 cells that project small cilia into the salivary mucus covering the oral cavity. We distinguish only five basic taste modalities: sweet, bitter, sour, salty, and umami (the specific taste elicited by monosodium glutamate), even though the molecular diversity of tastants is extreme, ranging from small ions to very large proteins. Thus, in contrast to smell, taste detects but does not discriminate between a large variety of molecules (Dulac, 2000). Because of this diversity, several pathways are thought to be involved in the reception and transduction of tastants, either via binding to receptor proteins (bitterness, sweetness, umami) or by interacting with ion channels in the taste receptor cell membrane (saltiness, acidity) (Laing and Jinks, 1996). The understanding of taste coding and information processing is still very limited, but the recent identification of a large family of 40–80 bitter taste receptors opens the way to significant progress on that topic (Adler et al., 2000). Chemesthesis In addition to taste and smell, there is a more generalised chemical sensitivity in the nose and mouth mediated by the trigeminal nerve, which is designated by the term chemesthesis (Green and Lawless, 1991). Comparatively less experimental effort has been expended in understanding the transduction processes involved in this sensory system, although the presence of several receptor mechanisms, ranging from receptorspecific to non-specific, has been hypothesised (Brand and Bryant, 1994). A variety of flavour experiences arise from trigeminal stimulation: the fizzy tingle from carbon dioxide in soda, the burn from hot peppers, pungency from black pepper or other spices, the cooling effect of peppermint, or the nasal pungency of mustard and vinegar (Lawless and Heymann, 1998). Qualitative distinctions among chemesthetic sensations are not easy, either due to lack of vocabulary or to poor discriminative ability. Another confusing element is that some irritants, e.g. menthol, also have taste or odour properties, while some tastants or odourants, like salt or acetic acid, can produce irritation (Prescott, 1999). 2.2.4 Interactions An important aspect of sensory perception in the context of this book is that foods generate a multitude of stimuli, which are processed more or less simultaneously to produce an integrated, overall impression. Along the way from the food item to the brain response, there are many possibilities for stimuli (particularly chemical ones) or sensory perceptions to interact, resulting in a response that is not simply the sum of individual responses to each independent stimulus. Interactions between stimuli within the same modality have been extensively studied for tastants (see Breslin (1996) and Birch (1996) for reviews), and also for odourants. Inter-modality interactions have been reported, particularly between irritants and tastants, and between irritants and odourants (Lawless and Heymann, 1998). Whether these interactions are physiological (at the receptor site) or psychological (central processing) remains unclear. There is no evidence for a physiological interplay between taste and

36

Instrumentation and sensors for the food industry

olfaction (Laing and Jinks, 1996), but under normal conditions of real food consumption, there is certainly a cognitive interaction between taste and aroma (Noble, 1996), not to mention the additional influences of texture and appearance.

2.3

Sensory evaluation methods

2.3.1 Discriminative tests Discriminative tests are used to determine whether or not a difference exists among samples. Two or several samples can be compared. Several types of discriminative test exist, each one corresponding to a particular situation. All are ‘forced-choice’ methods. The Triangle test is probably the most universal sensory method. It is a nondirectional difference test, used to detect an unspecified sensory difference between two samples. The panellist is presented with three coded samples: two samples are the same and one is different. He is asked to identify the odd sample, optionally he may comment on the intensity and type of difference. This test is useful in quality control, to check samples from different production batches, and to determine if any change of ingredient or process induces a detectable difference in the product, but it should be limited to relatively homogeneous products. The Duo-trio test is also used to detect unspecified differences between two samples. Three samples are presented: one is labelled as reference and two are coded samples, one being the same as the reference and the other being different. The panellist is asked to identify which of the coded samples matches the reference. This test, similar to the triangle test in its objectives and uses, has the advantage of limiting the number of comparisons (i.e. it is well suited for intensely flavoured products) but it is less powerful due to a higher probability of selecting the correct sample by chance. The above two methods do not require any specific training of the panellists. The Pair comparison test is a directional difference test, used to detect a specified sensory difference between two samples. The panellist is presented with a pair of coded samples and is asked to determine which one is more intense in a particular characteristic. This test is very simple, often used in quality control but its efficacy depends on the relevance of the chosen criterion of comparison. The Ranking test is an extension of the pair comparison test. The panellist is presented simultaneously with a series of coded samples (3–10) and is asked to rank them in order of intensity for a single specific attribute. This test is generally used to screen samples from a large group. It does not give any indication about the magnitude of the difference but only an ordinal evaluation, which is why results from different ranking tests cannot be compared. When these methods are used to assess products, panellists should be selected according to their sensitivity to the attribute of interest. These two methods are also well suited to select and train panellists for descriptive tests. Further to discriminative tests, the sensory analyst is generally interested in obtaining more detailed information about sensory characteristics of the products: descriptive tests aim to identify, measure and compare the sensory attributes of the products. 2.3.2 Descriptive tests The aim of a descriptive test is to describe with a complete and non-redundant set of pertinent attributes the sensory properties of a product and to quantify the perceived intensity of each attribute on a scale. The outcome of this process is referred to as a

Instrumental measurements and sensory parameters

37

sensory profile. Several types of descriptive tests have been developed. The Flavor profile was the first technique developed (Arthur D. Little Inc., 1958). A small panel of four to six trained assessors analyses and discusses the flavour characteristics in both qualitative and quantitative terms. The products are first analysed individually for aroma, flavour, mouthfeel, after-taste and overall impression or ‘amplitude’, then the assessors sit around a table and discuss their evaluation in group to reach a consensus. Extensive use of reference materials is made to achieve a complete description of the product. The final profile, consolidated by the panel leader, presents the detectable attributes (‘character notes’) in order of detection and their intensity scored on the following scale: 0 = not present )( = threshold 1 or + = slight 2 or ++ = moderate 3 or +++ = strong The final flavour profile being a group consensus, no statistical analysis can be carried out to compare different products. The Texture profileÕ was developed in the 1960s by the General Food Research Center. It is defined as ‘the sensory analysis of the texture complex of a food in terms of its mechanical, geometrical, fat and moisture characteristics, the degree of each present and the order in which they appear from first bite to complete mastication’ (Brandt et al., 1963). The Texture profile is rather similar to a flavour profile approach with panellists discussing results in order to reach a consensus. Each texture attribute is illustrated by references to represent each scale category. The number of scale categories can vary from five to nine, depending on the specific attribute. The reference foods are standardised with respect to brand name, handling procedure, sample size and temperature. The problem of such a method can be the variability of the references and the sensory fatigue due to extensive use of references during the test. Moreover, it is not realistic to separate texture from all other sensory properties because perceptions are interdependent and the exclusion of some other sensory attributes from the profile does not eliminate their perception. The Quantitative Descriptive AnalysisÕ (QDA) was developed by Stone et al. (1974) and is now widely accepted as the standard profiling method by most sensory analysts. It is a time-consuming and complex method but has broad applications such as description of complex products, comparison of products with identification and quantification of their differences, explanation of consumers’ preferences, correlation with instrumental measures. Thus QDA is widely used in product development, quality assurance, shelf life testing, cost optimisation and market survey. Results can be related to instrumental, chemical or physical properties of the products. The set up of this method requires several steps: • • • •

select and train a panel generate the descriptive vocabulary prepare the questionnaire and the glossary create an experimental design taking into account the number of products, number of tasters and number of replicates • organise the evaluation (sample coding, preparation, service, data acquisition) • analyse the results and present them in a synthetic and understandable form.

38

Instrumentation and sensors for the food industry

The ideal number of panellists is 10 to 12. This number is large enough to avoid too much variability of the results dependent on too few subjects, and small enough to ensure an efficient and homogeneous training of the group. Prior to training, the panellists should be screened according to their familiarity with the type of product considered, to their ability to perceive differences, to verbalise and to work as a group. After selection of the actual range of products to be profiled, the panellists are asked to generate a list of attributes to describe all the products’ sensory properties, grouped by modalities: appearance, aroma, flavour, texture, mouthfeel and after-taste. Sometimes only some sensory aspects are considered, for instance colour and appearance of the product, in which case it is called Appearance Profile Analysis (APA) (Hutchings, 1994). The generation of vocabulary is facilitated by the panel leader who helps in reducing the number of attributes, reaching consensus on the meaning of terms and writing the definitions of the glossary. Usually 100 to 200 descriptors are generated by the panellists, then the vocabulary is reduced to 15–30 key attributes in the final questionnaire. The role of the panel leader is to facilitate the discussion and to organise sessions, but he must refrain from influencing the group. The panellists are then trained with product or ingredient references to illustrate the different sensory characteristics. This training helps them to identify and memorise the attributes and to develop a standardised evaluation procedure. The setting up of the questionnaire includes the choice of a scale to rate the intensity of each attribute. The scale must be homogeneous for the whole questionnaire. Two main techniques are commonly used: • Absolute profiling: the attributes are scored on a structured (numbered) or an unstructured (visual analogue) scale, generally anchored at both extremities by words. The scale is always organised so that the intensity increases from left to right. A 10point scale or a 15-cm continuous/linear scale is most often used. • Relative profiling: the attributes are scored in comparison to a reference sample situated in the middle of the scale. In this case the scale is organised so that the intensity goes from ‘much less’ on the left to ‘much more’ on the right. The relative scale is generally numbered from 5 to +5, with reference at 0. The relative scale is often used to assess deviations from the standard in QA tests or storage tests.

Training is essential to use the scale in a consistent and repeatable way. This is achieved by presenting several times standard samples illustrating the lowest and highest scores for each attribute, and intermediate levels. The complete training of a QDA panel is rather long; about ten 90-minutes training sessions are required to reach reliable and repeatable assessments. Nevertheless, and despite training, the panellists can have different perceptions and a certain amount of variability between them is unavoidable. The experimental design has to take into account: • the total number of products • the number of products presented within a session (4 to 6, depending on the complexity of the products) • the number of replicates (2 to 3) • the number and availability of tasters • the amount of time allowed to run the study.

To fulfil these requirements, products can be analysed according to a complete block design (all tasters analyse all products) or an incomplete block design (each taster analyses only some products). In any case the products must be evaluated independently and blind, with a monadic presentation (or against a reference for comparative profiling),

Instrumental measurements and sensory parameters

39

according to a balanced order of presentation to avoid bias due to order effect. Statistical treatment requires a complete dataset. Univariate and multivariate methods can be applied to analyse and visualise results: • The basic treatment consists of calculating the mean values and variability for each attribute and product. • Statistical differences between products can be assessed by a two-way analysis of variance (ANOVA) on each attribute. This analysis, completed by a multiple comparison test (Duncan, Newman-Keuls, LSD...) allows the identification of significantly different products. Results are visualised graphically by bar charts or profiles (linear or spider webs) as shown on Fig. 2.1. • In a second step, multivariate analysis such as Principal Components Analysis (PCA), Discriminant Analysis (DA) or Partial Least Square Regression (PLS) can be used to reduce the multidimensionality of the data and to look at the relationships between products and attributes. Results are visualised on a biplot graph. An example of PCA mapping of coffees is presented in Fig. 2.2.

Spectrum descriptive analysisTM was developed from the flavour and texture profile methods (Meilgaard et al., 1999). In this case, the panellist scores individually the perceived intensities with reference to pre-learned ‘absolute’ intensity scales in order to obtain profiles universally understandable and usable by everyone. The method provides lexicons of attributes similar to those of QDA; each attribute is illustrated by standards that define a scale of intensity going from 0 to 15. The use of this method requires extensive training, because panellists have to be universal. Free-Choice Profiling is an original technique of free description (Williams and Arnold, 1984). Panellists describe the products with their own words associated to their

Fig. 2.1

Aroma profile of two different soluble coffees presented in bar chart, linear profile and spider web.

40

Instrumentation and sensors for the food industry

Fig. 2.2

PCA mapping of eight soluble coffees with varying Arabica/Robusta blend, and varying roasting level.

own scales. The advantage of such a technique is that it saves time since no specific screening or training is needed for the panellists. This method can be applied to all sensory aspects of the products or limited to some characteristics i.e. flavour or texture. Six to ten assessors are generally required. Data are analysed using a Generalised Procrustes Analysis. Free-Choice profiling has been often used in consumer tests, it can be considered as a good communication tool to relate marketing or consumer profiles to traditional QDA. It can be applied in panellist training, for descriptors consensus within a panel, or for agreement between several panels. 2.3.3 Quality control tests The major application of sensory tests in production is to measure product conformity to a standard, in order to ensure quality consistency. A number of tests have been developed or adapted for quality control over the last 30 years, although some of the tests reported in the literature are not recommended for quality control purposes (Mun˜oz et al., 1992). Sound sensory tests for QC share some common and important features: • product standards and specifications based on some consumer input, defining what the product is expected to be and the range of acceptable sensory variability • a sensory panel trained on these specifications, using appropriate reference products • a test methodology allowing quick and unambiguous decisions regarding product release or corrective actions.

The Quality Ratings Method is derived from the early sensory approach relying on ‘experts’ to judge the quality of products, and uses a panel to evaluate daily production on a quality scale ranging from ‘very poor’ to ‘excellent’ (or equivalent designations). Products below a certain quality level or grade are rejected. This method has enjoyed vast

Instrumental measurements and sensory parameters

41

popularity, even outside the factory environment, and several professional associations have published such methodologies for various types of products (e.g. for dairy products, see Bodyfelt et al. (1988)). Despite its popularity, this method presents several methodological flaws and is recommended only under special conditions (Mun˜oz et al., 1992). Another widely used test is the In/Out Method (Nakayama and Wessman, 1979). A trained panel evaluates daily production as being either within (‘in’) or outside (‘out’) some well-defined sensory specifications. The percentage of panellists who score the product as ‘in’ determines the actions to take, according to decision rules set by the management. Advantages of this method are its simplicity, the direct use of the results and the panellists’ motivation to participate in the decision-making process. Disadvantages are that relatively many panellists are needed to be able to analyse data in a statistical way, and the lack of information on the source of the problem when a product is ‘out’. This can be alleviated by including comments or descriptive ratings in the test, but at the expense of its simplicity and speed. The Difference-from-Control Method (Aust et al., 1985) involves rating on a scale how much different a sample is from a designated reference. Products are rejected if they are beyond a cut-off point on the scale. This method requires a consistent easy-to-hold or easy-to-reproduce control. Like the In/Out method, it is fast to implement and operate, and panel training is relatively easy. However, the test result indicates the amplitude of the difference, not its nature. Here too, adding to the test some attributes rating provides additional information, but that increases also the time needed to train the panel and to execute the test. The most comprehensive approach is the Descriptive Method (Sidel et al., 1983). A well-trained sensory panel evaluates the intensity of several key sensory attributes, known to vary during production and to affect consumer acceptance. Specifications consist of tolerance ranges for each attribute, and a product falling outside any of these ranges is deemed unacceptable. This method provides the most detailed information about what is varying in a product, by how much and in which direction. Obviously, this is also the method requiring the most time for panel training and operation.

2.4

Sensory-instrumental relations

2.4.1 Principles There are frequent occasions when it is desirable to relate sensory and instrumental data. This can be to understand the chemical and physical causes of the sensory properties in a product, or to reduce the sample load on a factory sensory panel by replacing some of the tests with instrumental analyses. Needs will differ in research and in production environments, as will the approaches and means used to address them. In any case, this will require a thoroughly planned and executed ‘sensory-instrumental relation’ study. Definition of the objective The first step is to define clearly and fully the objective of the relation study, i.e. which sensory property(ies) has to be related with which instrumental parameter(s) for which product (type), and to design the experiments to be conducted in view of this objective. This is particularly important with respect to the range and distribution of samples that will be considered (see below). In a factory, a typical objective of such a study is to test instrumental methods that could be used to monitor product sensory quality (e.g., a micrometer for chocolate

42

Instrumentation and sensors for the food industry

grittiness, or a texture analyser for yoghurt firmness). Another goal can be to measure the physico-chemical characteristics of reference products used in sensory monitoring tests, to prevent them from drifting when they have to be renewed. A particular requirement in that environment is to employ affordable, robust instruments usable by unskilled personnel. In a research context, relation studies are used to investigate the influence of the physico-chemical properties of a product or ingredient on its sensory properties, or to assess the effect of processing parameters (including storage) on them. This may require more sophisticated instruments, but on the other hand, robustness and operator training are not limiting factors. A few examples of research objectives are listed below: • Assess the influence of process parameters (temperature, pressure, energy input, time, etc.) or equipment (size, supplier, etc.) on the sensory properties of a product. • Optimise the processing conditions to achieve a given intensity of a sensory attribute, or to keep it within certain limits. • Explore the potential of a different technology, or test new processing equipment. • Find chemical ‘markers’ to follow the evolution of sensory properties with time (during process, during shelf life, under abuse). • Identify physico-chemical characteristics that are related to key sensory attributes, in terms of consumer acceptance. • Predict the final product sensory profile on the basis of the raw material chemical properties.

Experimental planning Sensory considerations The sample preparation conditions (temperature, type of water, etc.) must be standardised, and appropriate test parameters must be selected (scale, use of a reference, number of replicates, etc.). A balanced presentation design is necessary to eliminate firstposition and carry-over effects. The performance level of the panel must be checked before the study. The trust in sensory data can be reinforced by showing its relationship with analytical data, but the panel performance cannot be assessed by any such relationship, particularly because the sample presentation design used for panellists assessment is different from the one used for samples evaluation. As for any sensory evaluation test, it is essential to ensure that samples are safe (microbiology, contaminants) before serving them to the panel. Instrumental considerations Forethought in method selection can considerably reduce the complexity of subsequent data analysis. The primary criterion for the instrumental method selection is its potential to be related to the sensory attributes identified as important. If several methods exist, only the most efficient should be selected. It is sometimes erroneously assumed that increasing the number of different characterisations reduces variability and increases the amount of information. More data is only potentially more information, but certainly more confusion (and more work). Adding more and more hay to a haystack will only increase the difficulty of looking for the needle in it. In statistical terms, this is referred to as the ‘curse of dimensionality’. Prior method validation is needed to ensure that the performance characteristics of the method meet the requirements of the intended analytical application. The potential of an analytical method can be reinforced by showing its relationship with sensory evaluation,

Instrumental measurements and sensory parameters

43

but its validity can not be assessed by any such relationship (likewise, the absence of such a relationship does not prove that it is not valid). It is important to follow rigorous procedures during sample preparation and measurements. For instance, make sure the physical properties of interest (in relation to sensory properties) are not masked by the recent thermal or shear history of the sample. Statistical considerations It is generally easy to gather large datasets during factory production or product development, but most samples are not produced for the specific purpose of relating sensory data with instrumental characterisation. A sound experimental design is necessary to obtain a valid model for the sensory-instrumental relation. This implies the three following steps: 1. 2. 3.

Identify the controllable parameters that will be varied (usually, these are compositional or processing parameters). Define the experimental region by determining a range for each parameter that corresponds to its expected variation in real conditions. Cover the defined region as homogeneously as possible, by using appropriate experimental design techniques.

The fundamental question in experimental design is not ‘how many samples are needed?’ but ‘how homogeneously is the experimental region covered?’ Moreover, it is crucial that the sample subsets used for sensory and instrumental characterisations are identical, which requires that samples be produced under stable and controlled processing conditions, and in sufficient quantity for each characterisation technique. Finally, data from any characterisation technique, even the best-mastered one, are never exact. Any measurement has some variability, the amplitude of which has to be known and controlled. Data analysis All available knowledge on individual datasets, and on possible relationships between different datasets, has to be gathered before analysing results. This helps to summarise data correctly and to select the most appropriate statistical methods. Summarising data Although some characterisation techniques (e.g. pH) yield a single value for each measurement, most techniques have a multiple outcome, i.e. they give a series of data points for each measurement. To simplify analysis and to focus on meaningful information, it is desirable to summarise multiple outcomes (Table 2.3). Sometimes pretransformation of data has to be considered. For example, a chromatogram can be taken as it is or standardised according to the odour thresholds of known molecules. Relating data A relation is obtained mathematically by a model that expresses one dataset as a function of another dataset. If two sets of data are related by an existing and accepted deterministic model, no statistical method is needed. Unfortunately, deterministic models are generally not available and empirical model building must be applied, based on the actual observations. Purely empirical models are valid only within the explored experimental region and should not be extrapolated outside of it. Data visualisation and descriptive methods, as well as prior knowledge, can provide hypotheses on the nature of the underlying relations. For instance, if a relation between

44

Instrumentation and sensors for the food industry

Table 2.3

Data summarising for different types of multiple outcomes

Dataset type

Definition

Signal

A signal can often be summarised by a reduced number of parameters: maximum peak intensity, area under curve, asymptotic behaviour, perimeter of the curve (or the related fractal dimension). Distribution of the values of a single A distribution can often be attribute in several classes. summarised by a reduced number of Example: particle size distribution. parameters: mean, mode, percentiles, standard deviation. Often difficult to summarise Characterisation of a product consisting of many single attributes. because the characterisation is complete only when looking at all Examples: sensory profiles, attributes together. However, some chromatograms. profiles may be composed of subprofiles, like a sensory profile including appearance, odour, flavour and texture.

Distribution

Profile

Summarisation

Characterisation of the product as a function of one variable (e.g. time). Examples: texture analyser, gas sensors.

ice-cream texture and bubble size is investigated, the model is likely to contain the cube (volume) of the mean bubble radius. Knowledge-based models might have some predictive power outside the explored product range. The best approach with empirical model building is to start with a simple model and to increase its complexity only if it becomes more accurate, i.e. if the adjusted R 2 still increases. Finally, the chosen model should be validated. Validation consists of confronting the model – based on a given set of data (the so-called training set) – with another set of data (the validation set). It ensures that the model found not only applies to data from the current products, but also to data from any other product located in the previously defined experimental region. This check is essential whenever the aim of the model is to become a predictive tool. Method transfer If the sensory-analytical relation study aims at building knowledge on specific food products, ingredients or processes, its results will simply be applied within an existing project or prompt a new study. If however the purpose of the study is to set up an analytical method in production, additional steps must be taken to transfer the method to a factory. One frequent pitfall is that a relation study is made with very sophisticated instrumental methods that will never be reasonably transferred to the factory floor. If the objective of the study is to develop a quality-control tool (on-line or off-line), the constraints in terms of cost, complexity, and robustness must be considered in advance. Each quality-control method requires a reference measurement which frequently necessitates a more expensive, high-precision technique, which itself must be validated. This should be considered when establishing a method and its internal control plan. 2.4.2 Examples Colour and gloss in sugar panned confectionery sweets Together with colour, gloss is a key parameter of the aspect of sugar panned

Instrumental measurements and sensory parameters

45

confectionery sweets and an important factor driving acceptance for young children. For this reason, a study involving sensory analysis and instrumental measurement was conducted to investigate the effect of polishing on the visual appearance of this type of sweets. Sugar panned confectionery sweets consist of a chocolate centre covered with several layers of sugar. The top layers are coloured, then covered with wax and polished in a drum to give a glossy, homogeneous and regular surface. The aim of this study was to optimise the coating process in order to reach the optimum gloss with the minimum polishing time. Different production batches of sweets were sampled directly from a production line every five minutes during the polishing process: from 0 (no polish, no wax) to 20 minutes polishing with wax. These samples were analysed by sensory methods and instrumental measurement with a spectrophotometer. Sensory analysis A panel of 13 trained assessors participated in the study. The colour tests took place in standardised conditions, i.e. in individual viewing booths (Macbeth Spectralight), lit with an artificial daylight source D65. The sweets were analysed using the Appearance Profile Analysis method. Five attributes of colour and appearance were scored on a 10-point scale: chroma, lightness, shininess, colour homogeneity, and surface regularity. As it is difficult to score the colorimetric attributes of colour due to a poor colour memory, the panellists were provided with colour references to anchor the scales of chroma and lightness attributes (colour chips from Munsell Book of Color). A series of sweets of the same colour but with different polishing times were presented according to a randomised-balanced design to avoid bias due to presentation order. The same assessment was carried out independently for the different colours of sweets. Only the results of yellow and orange samples are reported here. Sensory analysis did not demonstrate any significant change in chroma nor degree of lightness with polishing time. However an important increase of shininess is observed at the beginning of the polishing process, i.e. up to about 10–12 minutes for the yellow and orange sweets. After this time the polishing process seems to become inefficient as no increase of shininess can be sensorially observed (Fig. 2.3). The homogeneity of the colour does not change with polishing, and the regular aspect of the surface is slightly increased up to 10 minutes polishing time. Instrumental analysis The colour measurement of sweets was performed with a ‘Macbeth Colour-eye 7000’ spectrophotometer equipped with a CIE standard daylight illumination source D65 (the same as for sensory analysis). A 38-mm view area was chosen. The presence of an adjustable specular port in this type of spectrophotometer allowed the measurement of the surface brilliance of samples by including or excluding the specular component (see Chapter 4 for more detail). Reflectance spectra obtained with specular component included (SCI) and specular component excluded (SCE) were different. Only the SCE mode showed differences due to gloss value. In fact, calculation showed that Lightness (L*) was the parameter most influenced by gloss effect, Lightness increased with gloss in SCE mode, corresponding to a lighter aspect of the colour, while no significant change in L* was measured in SCI mode. The difference between the SCI and SCE reflectance, i.e.
E* (SCI SCE), gave only the gloss value by excluding colour parameters. This gloss value could be related to polishing time and to the sensory evaluation of shininess.

46

Instrumentation and sensors for the food industry

Fig. 2.3

Increase of shininess as measured by sensory analysis; evaluation on three different batches of yellow and orange sweets.

The colour measurements on sugar panned confectionery sweets were performed with specular component included and an observation angle of 10º. The results presented are the average of 10 measurements. In general, parameters of hue and lightness were not affected by polishing time. Chroma tended to decrease, indicating that the sweets became greyer or less vivid with longer polishing time (Fig. 2.4). This decrease was not visually perceived, but it suggests that there could be a risk of colour alteration if polishing time is too long. For each product colour, gloss increased significantly up to 10–15 minutes polishing time, less afterwards. The sensory scores of shininess were very well related to instrumental ones expressed by
E* (SCI SCE) (Fig. 2.5). Conclusion This study shows that most of the visual parameters of panned confectionery: hue, chroma, lightness and gloss can be easily and effectively controlled by instrumental measurements. These methods could be applied for quality control in factories.

Instrumental measurements and sensory parameters

Fig. 2.4

47

Changes in instrumental hue, lightness and chroma with polishing time; measurement on three batches of yellow sweets.

However, some parameters of appearance such as colour homogeneity and surface regularity can only be examined by human vision, as they result from a multiplicity of stimuli synthesised by the brain as one piece of information. Sensory results, confirmed by instrumental ones, did not show any noticeable difference in colour and appearance between 15 and 20 minutes polishing, suggesting that the polishing time of this specific type of confectionery could be reduced to 15 minutes without affecting visual perception.

48

Instrumentation and sensors for the food industry

Fig. 2.5 Sensory-instrumental relationship for gloss; measurement on three batches of yellow and orange sweets.

Colour of UHT milk It is well known that fat content influences the perceived colour of milk. Conversely it has been demonstrated by several authors that visual parameters of milk or dairy products are used as cues for the evaluation of fat content (Tuorila, 1986; Pangborn and Dunkley, 1964). As there is a direct relationship between fat content perception and consumer preference, a change in milk appearance can influence acceptance. It is therefore very important to have a precise control of colour when developing new milk products, in order to maximise consumer liking. A study relating sensory and instrumental analyses of milk colour was conducted, first, to determine the visual tolerance of milk at two different fat levels and secondly, to assess how the colour and appearance attributes of milk vary with the fat level. The samples studied were prepared from two commercial UHT milks: skim milk (0% fat) and whole milk (3.8% fat). All intermediate fat levels were prepared by mixing the two milks in the relevant proportions. A panel of 12 assessors recruited according to their colour vision accuracy was trained for milk analysis using discriminating, ranking and scoring tests. The sensory assessments were performed in individual viewing booths lit with standard daylight source (D65, 300–830 nm, 6500K). The milks were presented in 30ml plastic cylindrical containers, filled up to the edges in order to have an easy view of the surface of the samples. The samples were observed according to standardised viewing geometry 0/45 (standard D1729-89 in ASTM, 1996) i.e., diffused incident light above the surface and viewing angle at 45º, with a distance of about 30 cm from the object. The

Instrumental measurements and sensory parameters Table 2.4

49

Fat levels presented in the duo-trio tests, for REF 0% fat and 2% fat respectively

REF 0% fat REF 2% fat

0.1%, 0.2%, 0.3%, 0.4% For lower limit: 0.6%, 0.8%, 1.0%, 1.2%, 1.4%, 1.6% For upper limit: 2.4%, 2.6%, 2.8%, 3.0%, 3.2%, 3.4%

instrumental measurements were performed using a spectrophotometer Macbeth 580, with 10-mm sample depth, D65 illumination and SCE mode to take into account the effects of surface aspect that can influence sensory judgement. Visual tolerance The aim was to define the visual differentiation threshold of UHT milks having different fat levels. Two tolerance areas were measured for 0% and 2% fat. The limits of differentiation were determined by duo-trio tests using respectively 0% fat and 2% fat as reference samples. Table 2.4 indicates the samples presented. The differentiation thresholds were determined at the concentration for which 75% of panellists gave the correct answer to the duo-trio test. They were calculated from the linear regression between percentage of correct answers (transformed into Probits) and level of fat content (Fig. 2.6). The reflectance curves of all milks were measured simultaneously by reflectance spectrophotometry to obtain the
E* corresponding to visual limits of discrimination. p
E
L2 ‡
a2 ‡
b2 The visual differentiation threshold for skim milk was calculated as 0.2% fat. This means that, up to 0.2% fat, milk is perceived as having the same colour as 0% fat milk. The corresponding
E* measured by spectrophotometry is 1.25
E* unit. Two

Fig. 2.6

Determination of visual differentiation thresholds: (a) from 0% fat content, (b) around 2% fat content.

50

Instrumentation and sensors for the food industry

Fig. 2.7

Delta E* values as measured from both 0% fat and 2% fat standards.

differentiation thresholds were calculated for low-fat milk (2%): 1.2% (lower value) and 3% (upper value). This means that the colour of a milk with a fat content of between 1.2% and 3% will be perceived as that of a 2% milk. The corresponding
E*s measured by spectrophotometry are 1.46
E* units and 1.25
E* units, as shown in Fig. 2.7. Sensory and instrumental measure of colour Milks of varying fat content were profiled using the Appearance Profile Analysis method already described (see previous example). At the same time they were measured by spectrophotometry using standard methodology. Both methods showed that most colour and appearance attributes are affected by fat content. There was good visual discrimination for the three main colour attributes: 1. 2. 3.

Hue changed from greenish to yellowish with increasing fat content. This change was particularly marked up to 2% fat. Chroma or colour saturation decreased as fat content increased, i.e. the green colour of low-fat milks seems more saturated than the yellow colour of high-fat milks. Lightness increased as fat content increased. This means that the green colour of low-fat milks appears darker than the yellow colour of high-fat milks.

The spectrophotometry measurements showed that the reflectance percentage (R%) increased across the whole visible range as fat content increased. This phenomenon is particularly marked at the higher end of the reflectance spectrum (Joshi, 2000). As discussed in Chapter 4, this is due to the higher scattering efficiency of fat droplets at longer wavelengths (Fig. 2.8). Figure 2.9 presents in parallel the changes of colour parameters with increasing fat content, as measured by sensory analysis and by spectrophotometry. Changes were similar and the correlation between sensory and instrumental data was good, particularly for hue and lightness. The correlation was also meaningful for chroma (R 2 ˆ 0.88), although results showed that increasing instrumental chroma corresponded to decreasing sensory chroma (Fig. 2.9c). This negative relationship can be explained by an increasing sensory perception of whiteness, whereas instrumental chroma measures the deviation from an ‘achromatic white’ towards colour, resulting in an increase in C* (instrumental indices exist to measure specifically whiteness, i.e. ‘CIE Whiteness Index’ mostly used in the paint industry).

Instrumental measurements and sensory parameters

Fig. 2.8

51

Reflectance spectra of milks with varying fat content.

Conclusion This study demonstrates that sensory analysis is generally a necessary prerequisite to define production standards. In this case, visual differences are determinant to fix tolerance area, and physical values (i.e.
E*) can be determined in relation to sensory data. This principle can be extended to other types of perceptions such as texture, taste or flavour. The instrumental methods developed to assess colour of UHT milks are very well related to sensory ones, so they can be used to control and optimise the colour of new milk products. Crispness of fried products Crispness is a desirable quality in many food systems, such as breakfast cereals, confectionery inclusions, savoury snacks, fried products, and so on. Much effort has been spent in academia and industry to develop an instrumental method to measure crispness, in order to understand and improve the crispness of products, and to ensure consistent high quality in production. These developments involved instruments that mimic the senses used to gauge crispness, i.e. tactile and acoustic. Typically, tactile information is obtained from the analysis of force-deformation curves of compressed samples. The acoustic part is analysed from recordings of sounds produced either by actually biting and chewing the product or by mechanical imitation. The following example describes an acousto-mechanical study on the crispness of coated fish products. Samples Five fish finger samples were produced in a pilot coating line with various breadcrumb materials. They were fried in a deep-fat fryer for better reproducibility and colour homogeneity. In order to obtain a broad range of crispness levels, a frying protocol was set up with two oil temperature settings (145º and 180º) and two frying times (3 minutes 10 seconds and 4 minutes 30 seconds). A total of ten samples varying in breadcrumb type, frying temperature and time were produced. Sensory evaluation A trained panel of ten people who had been selected according to their sensory abilities performed the sensory assessment. The training period consisted of 12 sessions of 90 minutes. In the first phase, the panel generated a profiling questionnaire focusing on appearance and texture, whereby panellists agreed upon the definitions and the evaluation

52

Instrumentation and sensors for the food industry

Fig. 2.9

Sensory and instrumental measurements of colour parameters: (a) hue (b) lightness (c) chroma.

Instrumental measurements and sensory parameters

53

procedures, and set the extremes of the scales. Then the panel was trained until its scoring was repeatable and consistent. The questionnaire included 13 sensory attributes describing the appearance (swollen, golden, crumb size, crumb homogeneity, fatty), the texture in hand (crumbly), and the texture in mouth (breading: hard, crispy, fat; fish flesh: juicy, chewy, compact, stringy). Each attribute was scored on a 100-point continuous intensity scale anchored at both ends (not = 0; very = 100). All tasters evaluated products in three replications, according to a balanced experimental design with five samples per session. Samples were fried according to a rigorous procedure, kept in the basket for 30 seconds above the oil bath and served within 2 minutes to panellists, with randomised three-digit codes. Analysis of variance showed that the products differed significantly (p 0.008856 L* ˆ 903.3(Y/Yn)1/3 for Y/Yn < 0.008856 a* ˆ 500[(X/Xn)1/3 (Y/Yn)1/3] b* ˆ 200[(X/Xn)1/3 (Z/Zn)1/3] where Xn, Yn, Zn refer to the nominally white object colour stimulus. Newer scales for determining small differences (BSI 1988), as used in the textile industry, are now included in the software of some automatic colour-measuring spectrophotometers. For example, the CMC(l:c) colour difference equation, where weighting factors are applied to the CIELAB L*C*h* data, produces a more uniform colour acceptability scale. 3.4.3 Further terminology Colour terms can be divided into the subjective and the objective (Hunt 1978). The subjective, the psychosensorial, are brightness, lightness, hue, saturation, chroma and colourfulness. Colourfulness, a recently introduced term, is that aspect of visual sensation according to which an area appears to exhibit more or less chromatic colour. Although hue is easily understood as that attribute described by colour names – red, green, purple, etc. – the difference between saturation and chroma is less easily comprehended. Saturation is colourfulness judged in proportion to its brightness, whereas chroma is colourfulness relative to the brightness of its surroundings. A similar difference exists between lightness and brightness. Lightness is relative brightness. Lightness is unaffected by illumination level because it is the proportion of the light reflected, whereas the sensation of brightness increases with increase in illumination. The objective terms, the psychophysical, refer to the stimulus and are evaluated from spectral power distributions, the reflectance or transmittance of the object and observer response. They provide the basis for the psychometric qualities which correspond more nearly to those perceived. For CIELAB space the more important terms are lightness L*, hue h* ˆ tan1 (b*/a*) and chroma C* ˆ (a*2 + b*2)1/2. Total colour differences
E* can be expressed either as the coordinates of colour space or as the correlates of lightness, chroma and hue. Hence
E ˆ ‰…
L †2 ‡ …
a †2 ‡ …
b †2 Š1=2 or
E ˆ ‰…
L †2 ‡ …
C  †2 ‡ …
H  †2 Š1=2 where
H* is used rather than
h* because the latter is angular. For small colour differences away from the L* axis, if h* is expressed in degrees, then
H  ˆ C 
h …=180†

72

Instrumentation and sensors for the food industry

Table 3.1

Overview on colour description systems and notation

CIE system (1931) This is based on the imaginary positive primaries X, Y, Z (transformed from real red, green and blue functions of trichomatic stimulus detection which may contain negative values). In CIE space, colour is located (assuming defined illuminant and observer) by (Y, x, y), where Y x, y

luminous reflectance or transmittance (containing the entire lightness stimulus) chromaticity coordinates x ˆ X =…X ‡ Y ‡ Z† y ˆ Y =…X ‡ Y ‡ Z†

CIE space is not visually uniform. Hunter Lab System (1958) In Lab space, colour is defined by (L, a, b), where L correlate of lightness a, b red/green and yellow/blue opponent coordinate correlates L ˆ 10 Y1/2 a ˆ ‰17:5…1:02X Y †Š=Y 1=2 b ˆ ‰7:0…Y 0:847 Z†Š=Y 1=2 CIELAB system (1976) In CIELAB space, colour is defined by (L*, a*, b*), where L visually uniform lightness a, b visually uniform chromaticness coordinates L ˆ 116…Y =Yn †1=3 L ˆ 903:3…Y =Yn † a ˆ 500‰…X =X †n1=3 b ˆ 200‰…Y =Yn †1=3

16 for Y =Yn > 0:008856 for Y =Yn < 0:008856 …Y =Yn †1=3 Š …Z=Zn †1=3 Š

Where Xn ; Yn ; Zn are the values of X ; Y ; Z for the reference white. Further terms used are h ˆ tan 1 (b =a ) hue C  ˆ …a2 ‡ b2 †1=2 chroma Other terms, such as adaptation, appearance, vision colourfulness, saturation and brightness, are discussed in Sections 3.3.1, 3.3.2 and 3.4.3.

A comparison of the major colour scales with the associated terminology is given in Table 3.1.

3.5

Instrumentation

Since colour is a psychological phenomenon, its measurement is based on human colour perception. Hence, photoelectric instruments must be corrected for both lighting and human visual response, while visual techniques must use observers with ‘normal’ colour vision under defined lighting. Examples of direct visual assessment are colour atlases for broad definition of the location of colours in colour space, collections or sets of printed or painted coloured papers specific to products or processes, and visual matching instruments

Principles of colour measurement for food

73

which use coloured filters. Typical of the former are the Munsell and Swedish NCS atlases which are structured on uniform colour space, and the Pantone collections of printer’s colours with defined ink mixtures printed from 10 to 100 per cent tinting strength. Probably the best known of the visual matching instruments is the Lovibond tintometer in which the object, under specified illumination, is viewed and matched against a series of coloured filters interposed over a white background by the observer. Photoelectric colour-measuring instruments can be divided into two classes, trichromatic colorimeters and spectrophotometers. The most successful of the early trichromatic colorimeters was developed in the 1940s by Hunter (1958). It comprised a light source and three wideband red, green and blue filters to approximate CIE standard illuminant C and the 2º observer. The tristimulus values obtained were transformed into Hunter L, a, b colour space. Until the advent of the computer and the photodiode such instruments were much less expensive than spectrophotometers and, although absolute accuracy may have been poor, they were very good at measuring the small colour differences demanded for industrial process control (Patterson 1987). The more modern tristimulus instruments are linked to computers with automatic calibration and the provision of a number of colour spaces. Such instruments may be supplied with a selection of sensor heads of different illuminating geometries to allow measurement of a wide range of product types depending on the nature and dimensions of their surfaces. A range of hand-held lightweight colorimeters and miniature diode array spectrophotometers, with optical geometries comparable in function with the larger bench instruments, is now manufactured by several companies. Their compactness is a direct result of the use of high energy pulsed xenon arc lamps combined with filtered silicon detectors and microchip circuitry. The comparative inexpensiveness of such instruments, with their built-in memories and choice of colour scales, has resulted in increased in-line colour measurement in all branches of industry where colour control is necessary or desirable. The most accurate colour-measuring instrument is the spectrophotometer. Reflectance instruments are usually fitted with an integrating sphere with the choice of including or excluding the specular component of reflectance. Care must be exercised in deciding which geometry is appropriate for particular applications. The diffuse component of reflectance from subsurface absorption and scatter is wavelength dependent, whereas the specular component is not. For materials with glossy surfaces the inclusion of the specular will increase measured reflectance which, when translated into colour space, can lead to large discrepancies in the interpretation of visual lightness, as usually viewed, and to a lesser extent of the chromaticness of the colour. For example, highly glossy black tiles used for instrument calibration have tristimulus Y values of approximately 0.3 when the specular is excluded but 4.5 when included. The consequence of this difference in Y of 4 per cent produces a specular excluded uniform lightness L* of 3 and an included L* of > 25. For medium grey and white tiles the excluded to included Y values are approximately 25 to 29 and 78 to 82 repectively, which give L* values of approximately 57 to 61 and 91 to 92 respectively. Hence the near constant effect of 4 per cent on Y from the specular reflectance produces a decreasing effect from black to white from > 20 to about 1 per cent in L*. The CIE recommends that colorimetric specifications of opaque materials should be obtained with one of the following conditions of illumination viewing geometries and should be specified in any report: • 45º/0 or 0º/45º, specular excluded • diffuse/0º or 0º/diffuse, specular included or excluded.

74

Instrumentation and sensors for the food industry

However, the spectrophotometers most commonly used for measuring colour do not have identical geometries. Three typical instruments have been compared by Patterson (1987), who points out that probably the biggest source of differences among the instruments can be traced to the specular component. Hunt (1987b) suggests that if measurements are to be compared it is better to include the specular because of the considerable variation in the area of gloss traps used in different spheres, but the more nearly correct measurements in relation to practical visual observation is with the specular excluded (Best 1987). For computer match prediction of pigmented materials, e.g. paint formulation, the total reflection (i.e. specular included) is preferred. This restriction does not usually apply to tristimulus colorimeters which normally exclude the specular component of reflectance where the illumination viewing geometry is 45º/0º, as in the classic Hunter bench colorimeter. Another important source of variation among tristimulus colorimeters and spectrophotometers is the area of the viewing aperture and the relative area of the illuminating light spot, which affects both the direction and the amount of light returned from translucent materials. MacDougall (1987) has demonstrated that translucent suspensions of milk exhibit a ten-fold decrease in K/S for an increase in aperture area from 5 to 20mm. Best (1987) states that accurate determination of K and S by measuring thin layers on black and white backgrounds requires that the ratio of the aperture area to the thickness of the sample must be considerably greater than 10, a criterion unlikely to be met for most foods. One further source of potential error, in addition to those associated with instrument geometry and sample structure, is the wavelength interval used to calculate the tristimulus values. Although the CIE (1986) specifies the standard observer at 5 nm intervals from 380 to 780 nm, such accuracy is not required for most practical purposes. For 10 nm accuracy the intermediate 10 nm values from the 5 nn tables should be used. However, the CIE has not yet officially recommended the use of 20 nm intervals, although many modern colour spectrophotometers detect at 20 nm intervals. Tables of weighting functions at 20 nm intervals for both the CIE illuminants and a variety of fluorescent lights have been calculated and are published in the up-to-date colour textbooks cited in this chapter. Errors attributable to wavelength interval are likely to be less important than those from instrument geometry, except when estimating the effects of narrowband emission lamps on materials with several absorption bands. Here the 20 nm interval may prove to be less effiecient.

3.6

Examples

The progress of pigment oxidation in fresh meat and the effects of illumination on orange juice are given as examples of the interaction of absorption and scatter on measured colour and visual appearance. The effect of the illuminant on the calculation of CIELAB from a variety of food specta is presented. 3.6.1 Fresh meat On exposure to air the purple ferrous haem pigment myoglobin on the surface of freshly cut meat oxygenates to the bright red covalent complex oxymyoglobin. During refrigerated display, oxymyoglobin oxidizes to brownish green metmyoglobin (MacDougall 1982; MacDougall and Powell 1997). Twenty per cent of surface oxymyoglobin with metmyoglobin causes the product to be rejected at retail because of its faded colour

Principles of colour measurement for food

75

(Hood and Riordan 1973). The changes in the mean reflectance spectra of over 100 packages of beef overwrapped with oxygen permeable film and held in the light at < 5C over a period of 1 week are shown in Fig. 3.5. As the pigment oxidizes there is an increase in reflectance in the green region of the spectrum as the alpha and beta absorption bands decrease and a distinct loss in reflectance in the red region with development of the metmyoglobin absorption band at 630 nm. The changes in colour, calculated for CIELAB and D65, are shown in Fig. 3.6. As meat fades there is a small loss in lightness L*, accompanied by much greater changes in the chromaticness coordinates a* and b*. The loss in a* and gain in b* can be interpreted as an increase in the hue angle h* in the direction of yellow; however, since it occurs with a concomitant loss in chroma C*, it is recognized as a more grey or dull colour. Dull yellow is perceived as brown. The appearance of meat is greatly affected by the colour rendering properties of the lamps used for display (Halstead 1978). Some fluorescent lamps recommended by the lamp industry for displaying meat have enhanced red emission which tend to maintain the preferred colour of oxymyoglobin and visually shifts the early stages of metmyoglobin development from brown towards red. This effect of red enhancement on meat colours has been shown to elicit a greater visual colour change in making brown appear red than in making red appear more red (MacDougall and Moncrieff 1988). For some people the flattering of red-enhanced lamps may make meat appear too red. The ICS Micro Match spectrophotometer used to measure these samples is equipped with the option of using alternative illuminants to calculate CIELAB. The estimated changes in meat colour attributable to different illuminants after 1 and 4 days’ exposure (Table 3.2) illustrate the effect that light quality has on lightness, hue and chroma. The changes in colour produced by the differences in colour rendering among some of the lamps are

Fig. 3.5 Reflectance spectra of fresh beef during oxidation of oxymyoglobin to metmyoglobin obtained on a diode array spectrophotometer at 20 nm intervals: means of over 100 samples wrapped in oxygen permeable film and stored at