Final Report Summary - EMPOSME (Enterprise Modelling and Performance Optimisation)
The primary objective of this project was to develop and implement an 'Enterprise modelling and performance optimisation' (EMPOSME) tool. EMPOSME integrates enterprise modelling and enterprise optimisation so that the decision maker can rapidly identify the risk / reward position of any choice he has to make vis-a-vis a suite of defined key performance indicators. EMPOSME enables practitioners to efficiently create an enterprise model and automatically transform this model into an enterprise optimiser that generates near optimal solutions to their required set of performance metrics.
The EMPOSME suite of tools, as validated through deployment at the SME end-user sites, is used as follows:
(1) The end user interacts with the enterprise modeller to describe (or model's) the entities in the organisation at enterprise level relevant to the main key performance indicator. The data for the supply-demand position is extracted real-time from the ERP system and / or legacy databases such as spreadsheets.
(2) The optimisation and planning module then plans multiple resources vis-a-vis a suite of key performance indicators. The plan is integrated, e.g. unlike standard planning systems the material optimisation does not assume infinite capacity and vice versa. The plans also allow for user interaction so that the practitioner can use his / her empirical experience to adjust the plan using visual interfaces to display the consequence of his / her decision on the relevant performance indicator.
(3) The performance manager shows the impact of the planning decisions taken today on the key performance indicator (predictive performance planning). For example, a series of visually interactive screens show the predictive performance of the enterprise based on the plan produced. The decision maker can thus rapidly identify the risk / reward position of any choice he has to make vis-a-vis a suite of defined key performance indicators.
The EMPOSME tool enables the experienced practitioner within an SME to accomplish the following:
- Rapidly model and accurately represent all of the entities (e.g. sales orders) and resources (e.g. people and machines) within the typical firm via a visually appealing user Interface.
- Enable the practitioner to define his multi-criteria enterprise objective function or goal.
- Enable the manager to rapidly diagnose and resolve the myriad of resource conflicts that emerge frequently within an enterprise.
- Enable the manager to evaluate the proposed solution and deploy it to people that complete the value generation activities within the firm.
- To develop and implement an integrated software solution to achieve an 'holistic optimised resource allocation solution' needed to process all sales orders in MTO and ETO industries for a near-optimum solution from initial quotation to return of the product at the end of its useful life.
The work in the first period focused on getting to prototype stage of the tool with one SME end-user partner:
1. capturing the main requirements from SMEs (WP2),
2. developing reference models from their process models, and proposing a new meta-model for a production modelling language (WP2 / WP3),
3. proposing a formalism to model the scheduling problems (WP3),
4. starting implementation of the solver prototype (WP4),
5. developing a portfolio of performance metrics (WP5),
6. deployment of the initial prototype at one end-user site (WP6), and
7. disseminating the knowledge through publications.
The second period focused on the generalisation of the EMPOSME tool and validation of the concepts developed during P1. This included specific work such as:
1. Extending the initial market survey to a wider representative of the industry, especially the development of a generic set of performance metrics to be used for optimisation (WP5).
2. Extending the meta-modelling concepts provided by UDE in WP3 to include the concepts of enterprise modelling for optimisation and performance modelling.
3. The automatically generated optimisers have been tested in production situations and it was found that by focusing on the core model elements there is significant commonality between the model structures from the individual end user firms. This presented the opportunity to develop a very powerful generic optimisation solver that performed well in all of the problem test sets.
4. Integration the enterprise model with the optimisation model for the general case.
5. Development of new generic scheduling model that describes the core of the production scheduling problem was developed. The development of a new generic model for scheduling problems with process and resource alternatives is the first time such a model has been formalised for scheduling problems.
6. Development of the solver to include several new filtering algorithms for constraints integrating reasoning on numerical variables (temporal relations) and logical variables (dependency constraints). These algorithms have been evolved through a series of iterations with a particular focus on the time of execution.
7. Additional functionality was added to the solver to validate the models built by the practitioners. In particular, algorithms were developed for discovering equivalence classes of activities, that is, activities that must be present (valid) together in the schedule. This is a part of more general pre-processing that discovers logical relations between activities. This pre-processor can also deduce some formal bugs in the model. This enables the manufacturing practitioner to build enterprise models, previously the domain of modelling and optimisation experts.
8. Development of a generic set of performance measures for optimisation that emerged as key priorities for SMEs in general.
9. Development of the performance manager to:
- first, identify and draw attention to possible future problems. Early identification of problems facilitates corrective action within the window of opportunity to do so.
- second, to make an optimal plan based on the current status in the production (orders, materials on hand, available resources, transport agreements etc).
- third, to show the impact of the planning decisions taken today on the key performance indicator (predictive performance planning).
10. Deployment and validation of the EMPOSME tool and concepts at the SME end-users.
11. Development of ERP5 to make the tool accessible to SMEs and configurable with the EMPOSME tool including:
- The 'business template' technology which provides a base abstraction of corporate information system, and
- The 'ERP5 express wizard' technology which automates the configuration of business templates to meet SMEs requirements.
At the end of this project the end user SMEs using the EMPOSME tool have achieved the following performance outcomes:
- sales order velocity increase of greater than 30 %;
- sales order on time in full (OTIF) delivery performance (as compared to original order promise date) increase to over 95 % from the current estimated level of 40 %;
- reduce planning time by greater than 60 % and planning effort by greater than 50 %.
The EMPOSME suite of tools, as validated through deployment at the SME end-user sites, is used as follows:
(1) The end user interacts with the enterprise modeller to describe (or model's) the entities in the organisation at enterprise level relevant to the main key performance indicator. The data for the supply-demand position is extracted real-time from the ERP system and / or legacy databases such as spreadsheets.
(2) The optimisation and planning module then plans multiple resources vis-a-vis a suite of key performance indicators. The plan is integrated, e.g. unlike standard planning systems the material optimisation does not assume infinite capacity and vice versa. The plans also allow for user interaction so that the practitioner can use his / her empirical experience to adjust the plan using visual interfaces to display the consequence of his / her decision on the relevant performance indicator.
(3) The performance manager shows the impact of the planning decisions taken today on the key performance indicator (predictive performance planning). For example, a series of visually interactive screens show the predictive performance of the enterprise based on the plan produced. The decision maker can thus rapidly identify the risk / reward position of any choice he has to make vis-a-vis a suite of defined key performance indicators.
The EMPOSME tool enables the experienced practitioner within an SME to accomplish the following:
- Rapidly model and accurately represent all of the entities (e.g. sales orders) and resources (e.g. people and machines) within the typical firm via a visually appealing user Interface.
- Enable the practitioner to define his multi-criteria enterprise objective function or goal.
- Enable the manager to rapidly diagnose and resolve the myriad of resource conflicts that emerge frequently within an enterprise.
- Enable the manager to evaluate the proposed solution and deploy it to people that complete the value generation activities within the firm.
- To develop and implement an integrated software solution to achieve an 'holistic optimised resource allocation solution' needed to process all sales orders in MTO and ETO industries for a near-optimum solution from initial quotation to return of the product at the end of its useful life.
The work in the first period focused on getting to prototype stage of the tool with one SME end-user partner:
1. capturing the main requirements from SMEs (WP2),
2. developing reference models from their process models, and proposing a new meta-model for a production modelling language (WP2 / WP3),
3. proposing a formalism to model the scheduling problems (WP3),
4. starting implementation of the solver prototype (WP4),
5. developing a portfolio of performance metrics (WP5),
6. deployment of the initial prototype at one end-user site (WP6), and
7. disseminating the knowledge through publications.
The second period focused on the generalisation of the EMPOSME tool and validation of the concepts developed during P1. This included specific work such as:
1. Extending the initial market survey to a wider representative of the industry, especially the development of a generic set of performance metrics to be used for optimisation (WP5).
2. Extending the meta-modelling concepts provided by UDE in WP3 to include the concepts of enterprise modelling for optimisation and performance modelling.
3. The automatically generated optimisers have been tested in production situations and it was found that by focusing on the core model elements there is significant commonality between the model structures from the individual end user firms. This presented the opportunity to develop a very powerful generic optimisation solver that performed well in all of the problem test sets.
4. Integration the enterprise model with the optimisation model for the general case.
5. Development of new generic scheduling model that describes the core of the production scheduling problem was developed. The development of a new generic model for scheduling problems with process and resource alternatives is the first time such a model has been formalised for scheduling problems.
6. Development of the solver to include several new filtering algorithms for constraints integrating reasoning on numerical variables (temporal relations) and logical variables (dependency constraints). These algorithms have been evolved through a series of iterations with a particular focus on the time of execution.
7. Additional functionality was added to the solver to validate the models built by the practitioners. In particular, algorithms were developed for discovering equivalence classes of activities, that is, activities that must be present (valid) together in the schedule. This is a part of more general pre-processing that discovers logical relations between activities. This pre-processor can also deduce some formal bugs in the model. This enables the manufacturing practitioner to build enterprise models, previously the domain of modelling and optimisation experts.
8. Development of a generic set of performance measures for optimisation that emerged as key priorities for SMEs in general.
9. Development of the performance manager to:
- first, identify and draw attention to possible future problems. Early identification of problems facilitates corrective action within the window of opportunity to do so.
- second, to make an optimal plan based on the current status in the production (orders, materials on hand, available resources, transport agreements etc).
- third, to show the impact of the planning decisions taken today on the key performance indicator (predictive performance planning).
10. Deployment and validation of the EMPOSME tool and concepts at the SME end-users.
11. Development of ERP5 to make the tool accessible to SMEs and configurable with the EMPOSME tool including:
- The 'business template' technology which provides a base abstraction of corporate information system, and
- The 'ERP5 express wizard' technology which automates the configuration of business templates to meet SMEs requirements.
At the end of this project the end user SMEs using the EMPOSME tool have achieved the following performance outcomes:
- sales order velocity increase of greater than 30 %;
- sales order on time in full (OTIF) delivery performance (as compared to original order promise date) increase to over 95 % from the current estimated level of 40 %;
- reduce planning time by greater than 60 % and planning effort by greater than 50 %.
