A Potted History of Infomace's Relationship
with the Australasian Dairy Industry
- VEHICLE SCHEDULING -
- LOGISTICS PLANNING -
- TRANSPORT MANAGEMENT SYSTEM -
OBJECTIVE
To improve facilities for planning and monitoring the efficient and effective farm collection and plant delivery of fresh milk throughout a Dairy Company’s organisation.
a) Introduction
This section has been prepared to provide a brief history and overall scope of the above projects, and Infomace International’s evolving Business Partnership with Australasia’s leading Dairy Companies, Bonlac Foods, Australia, and Anchor (NZDG) New Zealand.
These projects represent a significant acknowledgment of the use of the applied science of Computational Intelligence, in addition to traditional software, to enhance profitability and competitiveness.
As pioneers of Computational Intelligence the initial Infomace system created within the science, CIVS (Computational Intelligence Vehicle Scheduling) was implemented at East Tamaki Co-Operative Dairy Company (ETCDC) Ltd. in New Zealand and at Bonlac Foods Northern Region in Australia. It was developed for scheduling milk collections back to a single plant. The immediate effect was a fleet reduction in New Zealand and cost savings in Australia.
For larger clients or enterprise-wide consideration it was always realised however that a multi-plant environment needed to be addressed. This considerably increased the complexity of the solution, and increased the length of time required, not only to develop and tune the system, but also to allow the Computational Intelligence model to 'learn' the numerous options available to it.
In 1997 we completed the installation of a tailored model for Bonlac’s Eastern Region in Drouin together with a full Transport Management System interface, and for Bay Milk in New Zealand.
Concurrent to Vehicle Scheduling other Computational Intelligent Processing Control systems were also researched for both a new Bonlac milk processing at Darnum and existing sites including Cobden and Stanhope.
Over the last few years we have consistently worked with Bonlac Foods to implement associated systems such as CILOG (Computational Intelligence Logistics Planning) new ideas and further enhanced operational features to the transport systems we have provided. Indeed, we have just recently implemented Version 5 of the system throughout the Bonlac Foods organisation. This exercise is also currently being carried out for the New Zealand Dairy Group (Anchor).
So how does Infotram Work:
Amongst other things it looks at predictions of the milk volumes at every farm to be scheduled, (based on the previous collections), pumping time, the exact size of every vehicle used, and the exact lengths of every stretch of road that could be used. Many 'little rules' about roads have to be considered including the average speed, time to traverse them, situations whereby a vehicle needs a load before it can traverse them, and that some farm gateways can only be entered by vehicles travelling in one direction.
It then explores hundreds of thousands of possible ways to collect the milk which it expects to be at each farm, then, through the use of heuristics, narrows them down to the best possible options based on total kilometres, vehicle sizes and availability, running costs and various time constraints including avoiding excessive queues at cleaning and unloading bays.
So how do Schedulers work:
They look at previous volume collections and patterns in the same area, possibly maps with 'some' information on road lengths, sometimes need to rely on their memories for the 'little rules' road details, and finally adjust previous schedules to produce the optimal ones for the forthcoming runs.
The most important aspect of a scheduler's skill however is his broad experience and understanding of the overall situation. It is this aspect that most emulates a scheduler’s thought processes, yet paradoxically only provides the tools which will allow him to perform his job better.
INFOTRAM is able to explore more possibilities, which the human scheduler may never have considered. INFOTRAM is able to calculate exact schedule lengths and shortest paths quickly. INFOTRAM can more easily and comprehensively look at the 'total efficiency', rather than just the 'localised effect' of a schedule change.
Thus when an INFOTRAM Model, is first implemented it is but the birth of the system, and the beginning of the computational intelligence model's supervised learning, concurrent to the schedulers' own familiarisation in using the system.
INFOTRAM then has to learn from its environment and have its scheduling "skills" tuned via a fully integrated suite of 'front-end' facilities, while the schedulers will learn how to use them to their fullest capabilities AND even include INFOTRAM methodologies in what might be their own 'manual' considerations.
b) Background
§ CIVS was initially developed to use a Computational Intelligence model to produce a milk collection plan which includes associated vehicle distances and costs. It is now but part of the total INFOTRAM system.
§ The system takes a model of the collection/delivery area and calculates the best fit for selected vehicles with regard to time, distance and economy resulting in the optimisation of each collection/delivery.
§ This procedure is then integrated with comprehensive 'front-end' software that allows information to flow between INFOTRAM and a conventional data processing system. Over its life span the complexity of the features offered by the 'front-end' software has become extremely sophisticated.
§ A further development in INFOTRAM was to plan the departure and return of tankers during milk collection to avoid excessive queues for cleaning and unloading bays, and more significantly meeting hourly demand.
§
Subsequent
developments have included multi-plant, multi-product, factory diversions and
two kinds of split pickup.
c) Full Transport Management
Following up from the initial Infomace vehicle scheduling, Bonlac Foods Ltd. requested us to provide a full Transport Management System to additionally include:
§ Milk Supply Transport Reporting
§ Milk Supply Pickup Management and Reporting
Integration to:
§ Lab System
§ Supplier Payments System
§ Milk Collection Devices and Inputs
§ (Present and Future)
The system has been developed and implemented with the following main objectives.
(i) Similar in concept and operation to previous systems and procedures
(ii) Full integration to Computerised Vehicle Scheduling
(iii) Support new interfaces to Transport System
(iv) Support previous interfaces
(v) Support new Information Technology proposed for the future.
d) Goals and standard functionality of Infotram
To Make Substantial Cost Savings
While many intangible cost savings can be made including:
§ Elimination of errors in scheduling
§ Faster Reaction to changing milk collection needs
§ Opportunity for better product plant utilisation
§ Improved Schedules
§ Better Service to Suppliers
§ Improved Managing Reporting for decision-making
§ True Management-management of Schedules
§ Reduced Personnel Costs
§ Reduced Direct Tanker Costs
§ Provision for Multiple Products
§ Factory Diversion Capability
§ Explicit Washing and Fueling Facilities
§ Allowance for Multiple Vats
§ Partial Plant Drop-Off Facility
§ Accurate Milk Prediction
§ Split pickups
- the immediately tangible savings are in on-going Vehicle Running Costs.
TO CAPTURE THE EXPERT KNOWLEDGE OF THE PRESENT SCHEDULERS IN A SOFTWARE MODEL
The knowledge that staff gains through their employment with a company is often lost when they leave.
Although manuals are written, which invariably become out-of-date and training is given to the new staff, the expert knowledge is lost when the individual goes. In the case of a complex activity where skills have been developed over a number of years, this loss can prove to be at a large cost to the company.
TO PROVIDE SCHEDULERS WITH COMPUTER FACILITIES TO USE THE MODEL TO PLAN AND MONITOR MILK COLLECTION
By using the system a scheduler can quickly produce an optimised schedule.
TO RELEASE SCHEDULERS FROM COMPLEX AND TIME CONSUMING ACTIVITIES TO USE THEIR KNOWLEDGE TO FURTHER DEVELOP THE MILK COLLECTION SYSTEM IMPROVING THE SERVICE TO SUPPLIERS
Transport schedules can be developed in less than an hour rather than a number of hours with previous procedures.
With the time saved, schedulers can try different plans using different vehicles (both number and/or type) as well as providing management with information on the plan before the plan is implemented.
TO PROVIDE MANAGEMENT WITH UP-TO-DATE ACCURATE INFORMATION ON MILK COLLECTION
The system contains information on Farm Suppliers (including owner, location, distance from plant, off-road distance, vat size), Vehicle information (including capacities, weight, cost and efficiency information) and roading information covering road types, gradients and intersections.
The
system develops an optimised plan which includes vehicle running costs by
vehicle and by trip. This information provides an accurate financial
evaluation tool, on which significant management discussions can be made.
FURTHER COMPUTATIONAL INTELLIGENCE DEVELOPMENTS IN THE TRANSPORT ENVIRONMENT
Infotram evolved in 1997 and 1998 to provide the core system from which several other associated enhancement systems have been developed and again proven; it offers proven upgrade paths, connectability and integration features, and tailoring and management system enhancements.
These Include:
- Quarantine provisions (e.g. Anthrax)
- Milk supply prediction
§ PERFORMANCE MONITORING
- Key performance indicators
- Management reporting
- Cost comparisons
- Productivity comparisons
- Staff and contractor performance review
- Audits
- Ongoing quality control
§ MAPPING INTEGRATION
- Database
- Visual aid
- Animation
- Scheduling
- Training
- Planning
- Publicity
- Trouble-shooting
- Integration with scheduling
- Integration with supplier management
§ CENTRALISATION
- Database at a single location
- Schedule creation at a single location
- Despatching from a single location
- Automatic reassignment of farms between regions
- Automatic reassignment of trucks between regions
- Pursuit of company objectives, not regional objectives
-
Clear
responsibility and accountability
CILOG (COMPUTATIONAL INTELLIGENCE LOGISTICS)
§ CILOG PERFORMS SENIOR MANAGEMENT WHAT-IF ANALYSIS FOR STRATEGIC ENTERPRISE-WIDE DECISION-MAKING:
- Truck purchase
- Fleet deployment
- Shift structure
- Redistricting
- Takeovers
- Plant location
- Transport pricing
- Despatching strategy
- Negotiations with suppliers, customers and drivers
- Contingency planning
-
Risk management
e) REFERENCES
- ANCHOR MILK Group Transport Manager, Nevin Dwyer, March 2001