ControlDraw
Dedicated to improving your control systems with Process Control System models
Home
Company
About Control
Consulting
What's new
Our Clients
Library
ControlDraw
Overview
FAQ
Features
About Diagrams
Automation Models
Database
Screens
Support
Download
Updates
Online manual
Prices
Free Software
Other Downloads

Milkshake Plant Process Description

Overview Diagram

Additive Deliveries

On arrival of a lorry at the Additives reception bay the pallets of additives are unloaded by fork lift truck and taken to the Additive Warehouse.

A procedure is required to register the arrival, which should match an expected delivery as scheduled in the MES

Additive Warehouse

The Additive Warehouse stores palettes of Flavour additives. Each pallet contains a number of sacks or additives A system to control the location of materials and to ensure that they are used in an appropriate order, such as first-in first out or according to use by dates will be required. This document does not explore these areas of functionality.

Solids Processing

Flavour additives materials are depalletised then the sacks are slit, the contents extracted and then pneumatically conveyed into the Additive Silo.

Additives are then milled, sieved and then weighed into IBC's

This is driven by the production campaign requirements so that the right additives in the right quantities are processed.

Intermediate Product Store

The Intermediate Product Store provides a buffer which is used to store IBC's which have been prepared in advance of when they are needed or in case of a hold up in the liquids processing.

Milk Reception

On arrival of a tanker at the mile reception bay the tanker is connected up to one of the three possible unloading points. Milk is then pumped from the tanker into one of the 3 milk silos

On completion of the unloading of the tanker a clean is carried out of the tanker and the pipework which routes the milk from the tanker to the silos.

The milk silos operate on a continuous cycle whereby they are filled repeatedly during the day and then cleaned at night, when no milk deliveries are expected.

Only one tanker is unloaded at one time and generally to only one milk silo.

However, operators can in exceptions unload into one silo until it is full and then the next

CIP Chemical Deliveries

On arrival of a lorry at the CIP chemicals reception bay the drums are unloaded by fork lift truck and taken to the CIP Chemical Store.

A procedure is required to register the arrival, which should match an expected delivery as scheduled in the MES

CIP Chemical Store

The CIP Chemical Store stores drums of cleaning chemicals

 A system to control the location of the drums and to ensure that they are used in an appropriate order, such as first-in first out or according to use by dates will be required. This document does not explore these areas of functionality.

CIP Generation System

The Clean In Place  system provides rinse water and also doses the cleaning chemicals into the water to make cleaning liquids

The system also takes the returned liquids and can store the final rinse water for use in subsequent first rinses

Liquids Processing Building

Batch Mixing Processes are carried out to make batches of the various products. Certain products are then homogenised on their way to the Product Silos.

A typical mixing process comprises

Add Milk Start qty

Add Milk and Additive

Add Final Milk Qty

Mix

Mix and Heat

Transfer Out

 

Packaging Building

In the packaging building liquid product from one of the product silo’s is filled into botles and then labelled and packed into pallets - further details are given below.

The operation of the packaging lines is organised in terms of Work Orders.

A Work Order is characterised by:

The number of bottles to be packed

The labelling on the bottle - one work order per label type

The customer - a single customer per work order

A single work orde may be made up from one or more Product Silo

Bottles

Bottles are supplied to the filling line in boxes of pallets. A depalletising machine takes the boxes from each pallet, and the de-boxer takes bottles from the boxes. Bottles are then fed onto an accumation conveyor which feeds into the filling machine.

 

Customers

Milko Ltds supplies to a large number of customers including the premium Supermarkets, Milko Ltds own brand and some others.  Milko also has a strategy to be able to supply new customers in response to order placed over the phone or the Internet when the customers usual supplier is unable to meet the demand or has a production problem. The pricing of such orders is a strategic decision made by Milko’s senior management -  sometimes the cost will be high in order to make sure that the costs of the disruption to the schedule are more than covered, however sometimes the work is done at cost or even at a loss if the new customer is considered strategically important enough.

Rework

Just occasionally, and notwithstanding the highly effective scheduling system ,a batch is made which does not meet the critical quality standards of Milko Ltd.

This can  happen when there is an equipment failure - in general equipment failures can be tolerated by putting the process into hold, repairing the failed item, and then re-starting.

However, sometimes there are failures which cannot be fixed in time to allow the batch to be completed.

In these cases it is highly preferable to be able to re-use the batch if possible. This is far more acceptable than disposing of the batch down the effluent system, and can be achieved by using small percentages of the batch in subsequent batches of similar products provided this is done in time.

This practise is accepted to a greater or lesser degree by the customers.

The pipework in the plant makes it possible to divert a failed batch back into one of the milk silos for subsequent rework.

 

A further issue with rework is that when a problem does occur and some rework is needed , it can become  necessary to change the production schedule to make  product which can use up the reclaimed batch. (It may also require that the current schedule is altered in order to free space in a Milk Silo to hold the batch. )

Recipes

The plant produces a large number of products, in a variety of packages.

There are three basic types of recipe

Splodge Drinks

These include Banana, Strawberry and other fruit based milkshakes

Fudge Drinks

These include Chocolate and other cocoa based drinks

Alchoshake

These are the hot sellers, Milko Ltds latest introduction of milk shake alcoholic drinks have taken the club drinks market by storm since their introduction.

Recipe Characteristics

Each type of products is made in a different way

 

Within each general recipe type there are a large number of specific recipes. The Premium Supermarkets, and Milko themselves each have their own recipes but the lower price customers all use exactly the same recipe.

 

Rework is handled by a rework substitution algorithm - this is a set of recipe parameter which define how much of a particular type of rework batch can be added.  For example, small parts of banana milkshake can be added to all other product which also contain banana, no rework which contains alcohol can be used in any non-alcoholic products etc.

Packaging

The product is packed in a variety of ways. Each customer has their own labels for the bottles or cartons which are use. Alcoshake is provided in bottles with child-proof caps. And each customer has their own barcoding standard.

Operations

Homogenisation

Homogenisation is essentially a mixing process which ensures that the various constituents  of a batch are deeply mixed. The depth of the mix is determined by the physical properties of the batch and the pressure and flow through the homogeniser.

Homogenisation, if required, takes place in the process of transferring a batch from a Batch vessel to the Product Silos.

The process of starting up a homogeniser is a complicated by the mechanical problems of raising the pressure, which can be very high and the fact that it is possible to physically damage the equipment by running it empty. h

The homogeniser supplier provides a control system specifically for this equipment and refuses to guarantee it if this system is not used. Consequently Milko has bought the homogeniser and it’s control system as a complete package. The vendors system includes a serial interface so that all of the control parameters which the vendor considers relevant can be set from a supervising system

 


Milkshake Plant Issues

·       How do you break it into units?

·       What is the definition of a recipe in this instance?

·       What does a “batch” represent in this process?

·       How do we handle product traceability?

·       Can batches be split into sub-batches?

·       CIP - does this also constitute a recipe?

·       What procedures exist? If these are not recipe procedures then what are they?

·       How should transfer lines be handled?

How do the packaged batches conform to a recipe procedure.

   


Milkshake Production Plant Analysis

Building the Physical Model

It seems reasonable to start the analysis by building a physical model of the plant, as the P&ID should contain enough information on the equipment the plant is made of.

Building the physical model is the process of mapping the physical plant into objects which are found in the S88 physical model hierarchy

The following is a narrative of the thought processes which went on in doing this for the milkshake plant

Italics indicate questions raised

Underscore­ is used to indicate answers

Starting with the entire plant

Is this a process cell?

S88 says

process cell:  A logical grouping of equipment that includes the equipment required for production of one or more batches. It defines the span of logical control of one set of process equipment within an area.

Looks suitable so assume yes

 

Next, look at overall equipment content for obvious Units and Resources

 

Additive Warehouse

Is this a Common Resource

“common resource:  A resource that can provide services to more than one requester.”

Well I suppose a warehouse supplies a service of storing a material and providing that material on request so

Assume yes

(and the same for the other storage areas, Intermediate Storage , CIP Chemical Store and Packaging Store)

 

Building 1 Solids Processing

This produces IBC’s, which are themselves look like batches.

So is this a Process Cell

Assume yes

 

Therefore the whole plant is not one Process Cell, so what is it then

“area:  A component of a batch manufacturing site that is identified by physical, geographical, or logical segmentation within the site.”

OK let’s make the entire plant an Area

 

Milk Silos

Are these Units?

unit:  A collection of associated control modules and/or equipment modules and other process equipment in which one or more major processing activities can be conducted.

 

They look like units don’t they?

Well, yes, but they do not process batches as such. They just store milk.

So lets call them Common Resources then.

What about batch numbers?

Why?

We want to trace all delivered milk through to the product. So the batch log should identify the incoming milk batch number. In fact to avoid confusion we shall identify the milk by Lot number, not batch number.

So now we have a Common Resource with a batch (lot) number.

 

Liquids Processing Building

This is clearly a ‘proper’ batch processing area.

 

Summary - Plant Breakdown by S88 class - top level

 

Object

S88 Class

Additive Warehouse

Common Resource

Solids Processing Bldng

Process Cell

Intermediate Store

Common Resource

Milk Silos

Common Resource

CIP Chemical Store

Common Resource

CIP Generation System

Common Resource

Liquids Processing Bldng

Process Cell

Packaging Bldng

Process Cell

Packaging Store

Common Resource

Product Silos

Common Resource

Splodge Drink Procedure

Recipe Procedure

Fudge Drink Procedure

Recipe Procedure

Alchoshake Procedure

Recipe Procedure

Quarantine Store

Common Resource

Final Product Store

Common Resource

 


Boundaries between objects and CIP

When looking at a unit such as the Batch Vessels, there are things around them (the inlet valves are a good example) which intuitively seem to belong to the units. Look a bit closer, in particular when CIP or complex routing systems are involved in the process and the boundaries become less clear. For example in our plant the 4 port valves can carry both process fluids and cleaning fluids. Whilst they are under the control of a cleaning program it seems equally reasonable that they should belong to the CIP system as to the Unit which is being cleaned.

It is possible to view CIP processes as Servers to the Process Units, so that when a unit wants to be cleaned it can call on the CIP Common Resource to carry out the clean. However this seems unnatural - the cleaner when cleaning must be completely in charge of the equipment, and I have found that the Unit-Centric approach which S88 seems to promote, if taken too far can result in overly complicated cleaning systems which may even be very expensive in software and operational costs.

Another possible way to handle these complexities is to allow multiple physical and procedural models to exist in the plant, each applying depending on what the particular module needs to do at the time.

This is at first a bit mind boggling until we can really think of a plant as a large collection of relativeley autonomous objects, each behaving according to their own perspective.

S88 OK?

Is there a possible development - addition or modifications - to S88 which would address these problems

Or does S88 in itself have sufficient flexibility to cater for all of the solutions within it’s present terms and models?

Maybe the answer is yes to both questions.

 

Do the S88 models restrict the options? Is it necessary to break S88 ‘Rules’ to make or finish your analysis.

 

It certainly seems necessary to invent - or buy - other rules which S88 says nothing about.

 

If so then there seems to be a need for some follow up which provides the nuts and bolts whilst remaining true to the principles of S88.

If not then something is needed to provide a similar level of quality outside the batch areas as S88 provides within them.

 


 


Building the first product recipe

Can you mix unit procedures for which the batch sizes are different (but ultimately contribute the the final batch of product within one Master Recipe?

A Master Recipe applies to a Process Cell and if the Cell contains a mixture of batch sizes and even mixes up the batches, do S88 models still apply? Or what needs to be done to make them apply?

 

There are several possible ways of reconciling these issues

Enforce a rule that one batch = one process cell, so making it necessary to split the plant down into several process cells

Be more loose about the recipe and allow mixed batch sizes - this however may be a problem when it comes to actually mapping your model into a batch product


Comments

The whole reason for starting the Working Group was that the S88 concepts seem to be so applicable to the whole Process Control domain rather than just the specific area of Batches.

However, in proactise when you try to apply S88 in detail it becomes apparent that it is very batch-Centric. OK, the lowest level of the physical model - Contro Modules - stand up well even in continuous processes - and indeed since these are state oriented devices it is no surprise, since continous control is also very state oriented.

When it comes to the procedural levels of control there are also very usefull analogues between Batch procedures and those which apply to the startup and shutdown sequences which a continous process involves.

However, the more detailed aspects of batch recipes are hard to reconcile with continuous plant operations.

This is most clearly illustrated by comparing the filling parts of the milkshake plant with the liquid processing area. In the following diagram both the Unit Procedure in the Batch mixer and the are shown alongside each other as a sequence of operations (or maybe phases)

 

At fist sight the two look similar, and from the point of view of a granule of product they are.

IE each ‘granule’ of the product goes through the sequence in time in both procedures. In the case of the Liquid Processing procedure the batch is processed by the following sequence

Add Milk Start Qty

Add Milk and Solids

Add Milk Final Qty

Mix

Heat

In the case of the Filling procedure the granules are processed by the following sequence

Fill

Capping

Labelling

Film wrap  Carton

Shrink Wrap Carton

Palletise

 

However they are very different from a control point of view.

Each operation in the Liquid Processing procedure is completed before the next one starts, and the control sequence must reflect this - a single thread of the control recipe steps through the procedure and is only in one operation (and in fact phase and then step) at one time.

In the Filling procedure, all of the steps are executing simultaneously

 

One issue in this respect is that the entities which are being processed are different as shown below

If the filling procedure is shown in respect of a batch it becomes one operation, ie

This is more consistent with the Liquid processing procedure. In this case what is a batch?

It could be a work order - such as pack for a specific customer order. But we seem to have lost some detail. So where does that detail - the Fill, Capping, Labelling etc sequence now belong? Certainly the product is still subject to the same sequence.

Could it be that it has moved into the physical model? This does not seem reasonable to me because from the product granule point of view the original representation is better, and if recipes are not about the product itself then what are they about?

And if operations are about physical and chemical transformation then the what do we do with the filling line, which is most especially about physical transformation in this example (but can eqully well be viewed in terms of chemical transformation when looking at continuous processes.)

proposition

Continuous processes are a sequence of operations distributed in space acting on granules of material

Batch processes are a sequence of operations distributed in time acting on batches of material

 

So given that we as control engineers have to make use of the tools provided by system suppliers how can we make use S88 ( and hence of the batch-centric products designed around S88) to handle the filling line? We certainly need to have a system which will be able to track production through all of it’s stages and practically every plant will have a mix like the one given in the milkshake plant.

A solution is to break down the plant into separate process cells at each break in the batch entity. This I suspect mean that there are going to be far more process cells than most people would at first suspect.

And it might make using S88 based recipe management product far more complex that originally intended, not to mention challenging the license pricing policies of the suppliers.

Another major issue which arises, and one which perhaps is the most important of all is the handling of batch relationships.

I am not aware of anything in S88 which is designed to support the way batches in the various process cells along the route to producing the final product are realted to each other. It is I believe possible to model and track these, using relational database technologies. But what does S88 do to help? Is this something which S95 is addressing?

I recently asked a member of the S88 and batch software developer, who was presenting a new batch package, how it handled filling lines. The answer was that it didn’t. In fact I think he was underestimating the product, as I hope is illustrated above.

 The Milk Delivery process is another area of interest. Inasmuch as milk is delivered in batches then it is batch oriented. But the incoming tanker loads are mixed one on top of another.