Contents Lean Construction Ireland Annual Book of Cases 2020 Case 8 33 In the analyse phase of the DMAIC process, graphical improvement tools are frequently utilised. The gathered improvement ideas were classified into categorical data based on which quadrant of the Impact versus Ease of implementation graph they fell. Categorical variables represent types of data which may be divided into groups. A graphical representation of the categorical data is shown as a bar graph (Figure 2), representing the percentage of improvement ideas by quadrant category. In the improvement phase, ideas and designs were tested for their improvement impact. Appropriate remedial actions were planned after the analysis phase. A weekly idea-sharing forum was established with representatives from all international sites. An improvement roadmap was generated based on the ranked Priority Number (PN) scored for each improvement idea, owners were assigned to each improvement idea, and implementation dates by site were tracked. Each site developed its own walk-in ideas weekly forum where staff could bring in improvement ideas to be added to the central repository, and have them assessed and ranked by priority number. The outcome was a ranked roadmap of improvement innovations for implementation at each international site. During this improvement phase of the project pilot, implementations were conducted for some of the highest-ranked projects to verify their impact. Starting in March 2020, and after implementing the above solutions, the number of improvement ideas generated each week was monitored and plotted as a run chart. The final stage of the DMAIC approach involves the use of control plans and charts. During the control phase, the multidisciplinary team planned to gather and monitor the number of new ideas being generated over time. Here, a run chart was used to monitor the number of new ideas generated each week, and upper and lower process limits were calculated and added to the new ideas run chart. This type of chart is referred to as a process behaviour chart. Process behaviour charts assist in identifying changes in process behaviour and allow them to be differentiated from natural variation in the process. Using a process behaviour chart allowed the identification of when the number of new ideas being generated was unnaturally low or high. This allowed program changes to be monitored for their impact on the number of new ideas being generated each week. The upper and lower process behaviours limits were calculated from using the following formula: Lower Natural Process Limit = Average -3 (Moving Range Average)/1.128 Upper Natural Process Limit = Average +3 (Moving Range Average)/1.128 The moving range is the absolute value of the difference between consecutive points on the number of ideas run chart. The average moving range represents the amount of variation in the idea generation process. Analyse Phase Control Phase Improve Phase Figure 1. Idea Priority Number Figure 2. Idea by Priority Quadrant Figure 3. Ideas by DPS Site