Updated 5 June 2002 1 of 36 - 3/3/2016 Case 7 - Engineered vs. Catalogued Made-to-Order Supply Chains: Pipe Supports used in Power Plants 7.10 Case 7 - Engineered vs. Catalogued Made-to-Order Supply Chains: Pipe Supports used in Power Plants ................................................................................................................... 1 7.10.1 Business Case ........................................................................................................... 1 7.10.2 Background ................................................................................................................ 1 7.10.2.1 Power Plant Construction ................................................................................... 1 7.10.2.2 Pipe Supports in Power Plants ........................................................................... 2 7.10.2.3 Types of Pipe Supports ...................................................................................... 2 7.10.2.4 Pipe Support Suppliers and Product Standardization ........................................ 3 7.10.2.5 Need for Lead Time Reduction in Pipe Support Supply Chains ........................ 4 7.10.2.6 Methodology for Collecting Case-study Data ..................................................... 5 7.10.3 Characterization of Current Supply-chain Practices .................................................. 6 7.10.3.1 Alternative Supply Chain Configurations ............................................................ 6 7.10.3.2 Description and Analysis of Five SC Configurations and Sixth Alternative ........ 8 7.10.3.3 Selection Among Alternative SC Configurations .............................................. 10 7.10.4 Metrics ..................................................................................................................... 11 7.10.5 Analysis of Various SC Lead Times and Batch Sizes ............................................. 14 7.10.6 Analysis of Value-added Time over Total Delivery Lead Time ................................ 15 7.10.6.1 Value Stream Map ............................................................................................ 15 7.10.6.2 Value-stream Analysis Results ......................................................................... 16 7.10.6.3 Causes of Non-value-added Time .................................................................... 17 7.10.7 Detailed Value Stream Map for Fabrication ............................................................. 18 7.10.8 Process Simulation Showing Impact of Contributors to Lead Time ........................ 19 7.10.8.1 Value of Using Simulation ................................................................................ 19 7.10.8.2 Simulation of Design Process for Pipe Supports .............................................. 20 7.10.8.3 Design Process Model in STROBOSCOPE ..................................................... 21 7.10.8.4 Implementation and Simulation of Supply Chain Tasks ................................... 23 7.10.8.5 Scenario 1: Deterministic Model with Batching ................................................ 23 7.10.8.6 Scenario 2: Probabilistic Model with Batching, Variability, and Multitasking .... 25 7.10.8.7 Summary of Simulation Findings ...................................................................... 29 7.10.9 Supply Chain Improvements .................................................................................... 30 7.10.9.1 SC Improvements Implemented to Date .......................................................... 30 7.10.9.2 On-going Improvement Efforts ......................................................................... 32 7.10.10 Conclusions .......................................................................................................... 34 7.10.10.1 Case Study Conclusions ................................................................................. 34 7.10.10.2 Improvement Areas worth Further Investigation ............................................. 34 1 Updated 5 June 2002 1 of 36 - 3/3/2016 7.10 Case 7 - Engineered vs. Catalogued Made-to-Order Supply Chains: Pipe Supports used in Power Plants 7.10.1 Business Case Many construction inefficiencies are due to supply chain (SC) problems that occur at the interface between processes, disciplines, or organizations (Vrijhoef and Koskela 2000). This case study illustrates such problems in the supply of pipe supports used in power plants. The key business issue is that pipe supports often arrive late at the construction site, even though-relatively speaking the majority of supports are straightforward to engineer and inexpensive to procure. The need therefore exists to reduce the lead time for designing, procuring, fabricating, and shipping pipe supports in order to avoid late arrivals at the site. This can be done by using the capabilities available by all SC participants. 7.10.2 Background 7.10.2.1 Power Plant Construction Power plant construction has experienced an unexpected boom in recent years. “Between 1999 and 2001, about 83,000 MW of new capacity has come on line in the U.S., adding nearly 10% to the generation base” (ENR 2001). Real and perceived needs for new power have resulted in strenuous demand being put on the order-to-delivery time of power plants, so that today’s projects have to be executed not in a fast-track, but in a ‘flashtrack’ mode (ENR 1997). Power plant projects are complex and require thousands of components, including major mechanical and electrical equipment, vessels, structural steel, pipe, pipe supports, instrumentation, valves, fittings, and so on. SCs managed to make individual and sets of components flow, but also coordinated so that they will come together and match where and when needed, are a ‘must’ for project success. For each power plant component, SC participants not only affect the effectiveness and efficiency of the corresponding SC(s), they could also introduce a variety of inefficiencies. Inefficiencies need to be identified and eliminated to better satisfy not only customer, but all SC participant needs as well as to achieve project goals, otherwise, they will impact project completion. 1 Updated 5 June 2002 7.10.2.2 2 of 36 - 3/3/2016 Pipe Supports in Power Plants Power plants are configurations of a fuel storage and combustion system; a foundation system; a structural system; a power-generation-, power-conversion-, and transmission system; and various piping systems. They include piping systems for high- and lowpressure steam, feed-water, hot and cold reheat, and other systems such as condensate and heat recovery steam generator systems (HRSG). In turn, these systems include pipe as well as pipe supports, valves, in-line instrumentation, etc. A pipe support is as an assembly of components that attaches to the pipe and transfers the pipe’s load to the building or another structure in a manner that will prevent excessive movement under static and dynamic conditions during plant startup, operation, and shutdown. Therefore, pipe supports represent the interface between the building and other structural systems and the various piping systems, which interact with the location of equipment and vessels in the plant. 7.10.2.3 Types of Pipe Supports Examples of pipe supports are ‘constants’ (labeled ‘A’ in Figure 1) and ‘variable springs’ (B), ‘dynamic supports’ (or ‘snubbers’) (D), ‘slide bearings’ (F), ‘isolated supports’ (G, H), and ‘pipe shoes’ (pieces of pipe that transfer gravity loads to a structure underneath the pipe). C C C C C G B B A B E C B A I C E D G C C C C H F C C B D C H D A H Figure 1: Example Pipe Supports (from Pipe Supports Limited Inc. http://www.pipesupports.com visited on 02/26/02) A pipe support has three main parts: (1) the device which itself is also called a ‘pipe 2 Updated 5 June 2002 3 of 36 - 3/3/2016 support’, (2) the attachments (labeled ‘E’ in Figure 1) used to connect the pipe support device with the building structural system, and (3) the complementary hardware (‘C’ = pipe clamps and ancillary equipment; ‘I’ = turnbuckle) that connects the pipe support device with the steel attachment. 7.10.2.4 Pipe Support Suppliers and Product Standardization Suppliers of pipe supports offer a large variety of products, ranging from the supports they commonly advertise in company-specific catalogs (e.g., catalogs from Piping Technology and Products Inc., Lisega, AAA Technology and Specialist Co., Anvil International Inc., Shaw, etc.) to custom designs. To our knowledge, pipe support suppliers can fabricate in their job shops more-or-less anything the customer wants, especially if the business deal is right. On occasion, fabricators will nevertheless refer a customer to a competing supplier for exceptional specialty supports for which the customer can get a better deal than the supplier’s resale. A limited extent of business diversification stems from some companies focusing on supplying to the petrochemical industry and others to the power plant industry; others focus on specialty supports such as those needed to suit cryogenic conditions. Even so, cyclic markets force many companies to serve several industries. Pipe supports from Grinnell (now part of Anvil International Inc.) have de-facto served as a domestic industry standard because this company has been around for the longest time and, early on, on its own worked toward defining a comprehensive set of supports. Companies such as AAA allude by their advertizing to the equivalence of their products to Grinnell’s in ‘or equal’ procurement specifications. Despite the fact that companies know which of their products can substitute for a competitor’s, no industry-wide standard for pipe supports in their multitude of project applications presently exists, neither in the US nor internationally. The petrochemical sector of the industrial construction industry appears to be furthest along in terms of standardization whereas the power plant sector appears to lag behind in this regard, but further research is in order to determine what has driven and enabled these different sectors to move at different speeds towards industry-wide product standardization. Support suppliers compete in the domestic and overseas market by offering different 3 Updated 5 June 2002 4 of 36 - 3/3/2016 products and services. One example is an overseas supplier who offers supports, more compact than those otherwise available in the US, that are favored especially in cases when plant congestion is an issue. Another example is a domestic supplier who by default galvanizes all hangers, unless the customer insists on having them painted; other supports are galvanized only if customers request it. A galvanized finish is known to have a higher life-cycle value than a painted finish, but galvanization tends generally to be more expensive and adds to the delivery lead time. Given this company’s specific geographic location and the presence of multiple galvanizers nearby, these costs are not exorbitant. This company benefits overall in that it can reduce its SCM effort, thanks to the product simplification achieved by standardizing on a higher quality finish, while delivering greater value at a marginal if any extra cost at all to its customers. Based on life cycle analysis, this cost would be offset. From a SC performance perspective, a wealth of product variety is a mixed blessing. Engineering firms value the freedom of being able to design what they consider to be the best solution given the project requirements and then procure what they design, but focusing on product design alone does not automatically result in a good delivery process. If each and every product to be supplied to a project is unique, then managing the SC is significantly more complex than it would be if products were more standardized and supplied to a project in multiples. Custom engineering of a product may be a ‘penny wise and pound foolish’ proposition when the SC is not designed to accommodate it. The movement towards ‘mass customization’ aims to trade off such product- against process performance choices. 7.10.2.5 Need for Lead Time Reduction in Pipe Support Supply Chains Relatively speaking, most pipe supports are inexpensive and require straightforward engineering when compared to the cost and amount of engineering going into other power plant systems. Up to about 20% of the total number of supports in a power plant are customized. Nevertheless, problems in supplying pipe supports of any kind can compromise the success of the overall project. The reality is that a piping system is not complete and ready for start-up testing and turnover unless all pipe supports are in place. The problem may start early on in the delivery process. Pipe support design requires 4 Updated 5 June 2002 5 of 36 - 3/3/2016 input regarding the plant requirements and load conditions; the design of the structural steel system; the location of mechanical equipment, vessels, and instrumentation; as well as the physical (e.g., diameter, material, routing) and system characteristics (e.g., operating temperature and pressure) of the pipe that connects them. Current practice often is to define these inputs first and to push pipe support design towards the end of the power plant design process. Since power plants nowadays are managed as fast-track or even flash-track projects, design and construction overlap. When support design gets done in a rush and at the last minute, the downstream SC may get strained. For example, failing to allow sufficient lead time to fabricate and supply pipe supports to the site, or failing to coordinate the delivery of supports with the delivery of pipe and other system components, can make it necessary for field workers to use temporary supports (e.g., chainfalls) so that they can make progress on pipe installation and circumvent erection delay, but such practices create later out of sequence rework in the field, affect on-site piping productivity, and ultimately may result in project delays and budget overruns. 7.10.2.6 Methodology for Collecting Case-study Data This case study was initiated with the support of Parsons E&C one company <confirm with Frank that it is OK to name his company>who is a team participant member in PT172, and others expressing an interest in it. Data regarding industry practices at large however, was collected by means of an extensive literature review and interviews conducted with tens of practitioners working for a range of engineer-procure-construct (EPC) firms and pipe support suppliers. The resulting findings therefore do not represent practices at any one company, instead, they characterize industry practices in a more general way. The literature and interface with other researchers and practitioners in the power plant industry, including several with a CII research track record and others who have been involved in developing PIP, failed to reveal any research specifically focused on the design, delivery, or installation of pipe supports. This is striking, especially given the number of studies conducted to date on piping and related productivity, and common complaints about late deliveries. The data presented in this study was obtained from various firms and pieced together 5 Updated 5 June 2002 6 of 36 - 3/3/2016 as needed because the scope of the SC in this case spans multiple organizational boundaries. The research findings therefore do not represent any one specific project, though they are indicative of current practices in the power plant construction industry. The researchers would like to challenge individual companies and their SC partners to consider the presented analysis and conclusions as a basis for repeating the research process using their own process and project data in order to confirm the orders of magnitude of the metrics and develop an appreciation for the research findings and SC opportunities presented here. 7.10.3 Characterization of Current Supply-chain Practices 7.10.3.1 Alternative Supply Chain Configurations To characterize the SC for the delivery of pipe supports, five alternative configurations have been captured in distinct cross-functional maps. Each map shows the activities performed in engineering, procurement, and fabrication. The main participants in the SC for pipe supports are (1) engineering firms, (2) pipe support suppliers (who detail and fabricate the supports), and (3) contractors. Pipe fabricators may play a role in this SC but they are not necessarily involved. Engineering (1) and contracting (3) may lie within the scope of work of a single EPC firm. Delivery to site and construction were not detailed in this case study. (Meaning intent not clear in last sentence). The five alternatives require more-or-less the same activities to design and fabricate the supports. Different SC participants, according to their business interests, competencies, and capacity, may take responsibility of the design and/or detailing of pipe supports in order to suit different project requirements. The configurations are: Configuration 1 (Figure 2): Engineering firm designs the pipe supports. Supplier details, fabricates, and supplies the supports. Contractor installs. This model describes, by far, the most common practice in the industry. Configuration 2 (Figure 3): Engineering firm routes pipe and performs pipe stress analysis. Supplier designs, details, fabricates, and supplies the supports. Contractor installs. 6 Updated 5 June 2002 7 of 36 - 3/3/2016 Configuration 3 (Figure 4): Supplier fully designs pipe supports. Contractor installs. Configuration 4 (Figure 5): Contractor takes responsibility for pipe-support design and fabrication, though likely will subcontract this work out, and then installs. Configuration 5 (Figure 6): Pipe Fabricator takes responsibility for pipe-support design and fabrication. Contractor installs. Figures 2 through 6 show abridged versions of these five supply chain configurations. Detailed versions are presented in Arbulu (2002) and Tommelein and Arbulu (2002). Route Pipe LocatePipe Supports Perform Pipe Stress Analysis Design Pipe Support Chec k Interference and Loads Select Supplier & Send Info Approve Analyze Information Negotiate Engineering Firm Fabricate Pipe Supports Deliver Pipe Support Pipe Support Supplier Matc h Pipe and Supports Ins tall Supports Contractor Figure 2: Configuration 1 - Engineering Firm Designs and Supplier Details, Fabricates, and Supplies Pipe Supports used in Power Plants Route Pipe Locate Pipe Supports Perform Pipe Stress Analysis Select Supplier Check Interference and Loads Send Inform ation to Supplier Approve Analyze Inform ation Negotiate Engineering Firm Design Pipe Support Fabricate Pipe Supports Deliver Pipe Support Pipe Support Supplier Match Pipe and Supports Install Supports Contractor Figure 3: Configuration 2 - Engineering Firm Routes Pipes and Performs Pipe Stress Analysis. Supplier Designs, Details, Fabricates, and Supplies the Supports 7 Updated 5 June 2002 8 of 36 - 3/3/2016 Select Supplier & Send Info Route Pipe Check Interference and Loads Approve Budget Design Pipe Support Negotiate Engineering Firm Analyze Inform ation Locate Pipe Supports Perform Pipe Stress Analysis Fabricate Pipe Supports Deliver Pipe Support Pipe Support Supplier Match Pipe and Supports Install Supports Contractor Figure 4: Configuration 3 - Supplier Fully Designs Pipe Supports Route Pipe Locate Pipe Supports Perform Pipe Stress Analysis Send Inform ation to Contractor Engineering Firm Fabricate Pipe Supports Deliver Pipe Support Buy from Supplier Match Pipe and Supports Pipe Support Supplier Analyze Inform ation Design Pipe Support Install Supports Contractor Figure 5: Configuration 4 - Contractor Takes Responsibility for Pipe-support Design and Fabrication Route Pipe Locate Pipe Supports Perform Pipe Stress Analysis Send Inform ation to Pipe Fabricator Engineering Firm Fabricate Pipe Supports Deliver Pipe Support Pipe Support Supplier Analyze Inform ation Des ign Pipe Support Buy from Supplier Pipe Fabricator Match Pipe and Supports Install Supports Contractor Figure 6: Configuration 5 - Pipe Fabricator Takes Responsibility for Pipe-support Design and Fabrication 7.10.3.2 Description and Analysis of Five SC Configurations and Sixth Alternative Configurations 1 through 5: Configuration 1 is most commonly used today for pipe support delivery. This is a ‘cascading’ configuration with more-or-less sequential 8 Updated 5 June 2002 9 of 36 - 3/3/2016 handoffs between organizations. The resultant cascading effect is an important characteristic in that it allows the information to flow throughout the SC activities with a low level of interdependency. Less interdependence between participants typically means better flow of information, although the flow may get stretched out and otherwise impeded. By contrast, configuration 2 shows greater reciprocal dependence between the engineering firm and the support supplier who is in charge of all pipe support design. Reciprocal dependence allows for greater concurrency and less rework but requires collaboration to be successful. Sometimes, engineering firms do not have sufficient inhouse capacity to engage in support design and therefore hire the supplier to provide this service. Similarly, configuration 3 shows reciprocal dependence between the engineering firm and the supplier, but here, the information exchange requirements are even greater than they are in configuration 2, because pipe stress analysis is more involved. Pipe stress analysis is a service offered by some support suppliers, but engineering firms appear to favor doing this work in-house for a variety of reasons, including respective system operation characteristics and plant performance issues Configurations 4 and 5 also cascade, but the contractor or the pipe fabricator, respectively, rather than the support supplier, receives the handoff from the engineering firm. These configurations have been used in practice for small supports where no significant engineering is required, based on the assumption that pipe fabricators do not have the skills necessary to design and fabricate supports. Configuration 6: A sixth configuration that reflects vertical integration between the engineering firm, the pipe support supplier, and the pipe fabricator has been identified but was not studied in detail. The Shaw Group (Shaw), initially known for fabrication of pipe, then for fabrication of pipe supports, and more recently in the news for its acquisition of Stone and Webster Engineering in 2000, represents this new configuration. Vertical integration across so many tiers of the SC to yield a single company enabling Shaw to compete heads-on with many of its own customers, which are EPC firms. Shaw has become a SC integrator by putting equity into firms that originally were upstream and downstream from their initial core, pipe fabrication business. They also 9 Updated 5 June 2002 10 of 36 - 3/3/2016 have invested heavily in IT and other technologies (e.g., pipe bending to replace costly welding operations) to expand the range as well as the product and process quality of their capabilities. Shaw experienced that gross profit margins in engineering and construction are lower than those in other parts of its business, but it also is “changing the status quo of running EPC businesses by conceding that engineering is a commodity and that engineering services must be sold with a value-added component” (ENR 2002). An integrated SC package appears to appeal to a variety of customers in the market for power today. 7.10.3.3 Selection Among Alternative SC Configurations The selection among alternative SC configurations to best suit any one project must take into account numerous factors, including the capabilities (e.g., core and non-core competencies), capacity, and strategic corporate goals of the companies involved, as well as industry trends and the current and forecast market situation. In practice, engineering firms may use more than one SC configuration to balance the needs of several, concurrent and prospective projects. For example, on one project, the engineering firm may select a supplier early and collaborate with them using configuration 2 for the engineered-to-order supports, then involve that supplier using configuration 1 for all remaining supports that simply can be selected from that supplier’s catalog and made to order. The selection of a SC configuration for pipe supports may be governed by decisions the power plant owner makes. In part due to the rapid growth of the power plant industry during the last few years and, accordingly, the number of projects individual owners wanted to initiate within a short time span, some owners have established direct alliances or long-term agreements with major equipment suppliers (also see the “Transformer” case in this research report) but also with pipe fabricators and even with pipe support suppliers. In order to improve their own business performance, EPC firms have also established alliances with support suppliers. For example, Electronic Data Interchange (EDI) initiatives have been implemented in order to ease and expedite the interfacing between processes. EDI initiatives represent a good foundation to support increasing levels of 10 Updated 5 June 2002 11 of 36 - 3/3/2016 standardization of products and processes as well as power plant modularization initiatives. To achieve the best results in terms of value (including cost, quality, delivery lead time, product standardization, etc.), the owner makes decisions about whether or not its supplier alliance or an EPC firm’s alliance offers the best solution for each particular power plant project. 7.10.4 Metrics To complement the SC maps, several metrics were selected to gauge performance in different supply chain phases as well as in the SC at large. Some data for metrics was readily available whereas other data was more difficult to obtain from practitioners’ current project management systems. Examples of metrics are the following: Lead time: Lead time may refer to the entire time elapsed from order to plant start-up, or more specifically to, e.g., the time needed to approve detailed drawings prior to the start of fabrication, the time to fabricate a support, the time to deliver supports to the site, staging time on site (arrival of supports prior to their installation), etc. Supply-chain lead time depends on various factors, many of which pertain to the complexity of a product and variety in a product line. Supply-chain lead time is the sum of five elements: (1) direct work or processing time, (2) inspection time, (3) wait time, (4) move time, and (5) decision-making time. Decision-making time may be critical especially when several participants interact, playing different roles for different organizations. Information processing may get delayed for days or even weeks, thereby holding up other SC steps. The lead time of a single SC is of relative importance when compared to the overall project delivery process. In order to cope with project complexity, project managers often rely on the 80/20 rule-of-thumb: focus on the 20% of the activities, components, or the like, that contribute to 80% of the cost, delays, or other problems. Using this rule, however, pipe supports seldom make it to their list of priorities. Pipe supports are not typically ‘critical’ or ‘pacing’ items in a schedule as for instance turbine-generators or transformers are. According to this kind of thinking, also reflected in the PEpC study (CII 11 Updated 5 June 2002 12 of 36 - 3/3/2016 PT-130 ref), pipe supports on the critical path would be a scheduling anomaly. Now consider the 80/20 rule after it has been successfully applied. The 20% has been managed so as to no longer be a problem. The criticality therefore shifts to one or several items that previously were in the 80% list. Management must therefore be flexible and redirect its attention as the project gets executed in order to recognize and deal with these shifts. This presents an example scenario when pipe supports are critical. In CPM scheduling, this phenomenon is well known. As a schedule increasingly gets compressed, more parallel paths in the network become critical (also see Goldratt’s “The Goal” that describes how bottlenecks shift in a production setting). Project managers on flash track projects therefore have more to manage than managers on more slowly paced projects had to. Lead time also is of absolute importance. Overall benefits of lead time compression are: (1) faster delivery of the product or service to the customer, (2) reduced need to accurately forecast future demand, (3) less opportunity for disruption in the SC due to (design) changes, (4) greater possibility that participants will interact in a timely fashion with other SC participants, (5) easier synchronization of one SC with others (e.g., merging supply chains at the site), and (6) less opportunity for products to become obsolete. It is possible to directly attack the most visible waste just by flowcharting the process, then pinpointing and measuring non-value-added activities (Koskela 1992). This brings us to the next metric. Value: SCM aims to deliver value to the end customer. Accordingly, all SC activities must be assessed in terms of what value they bring to a customer. This is not easy to do! Direct work time (as distinct from contributory work time and wasted time) may be used as a token for value, though this hides inefficiencies in process execution due to poor work methods design and in the allocation of work to those who could get it done in the best way (work structuring). After determining value vs. non-value adding activities, a lean transformation process will seek to eliminate all waste from the activities and operations with handoffs between them that are executed to bring that product (or service) to market. This set of activities, operations, and associated information make up the value stream. A value stream perspective should look across individual activities, functions, departments, and 12 Updated 5 June 2002 13 of 36 - 3/3/2016 organizations, and focus on system efficiency rather than local efficiency within any one of these. Value streams are mapped and analyzed using a tool known as Value Stream Mapping (VSM). VSM was created by practitioners at Toyota as a tool to “make sustainable progress in the war against muda” (‘muda’ is the Japanese word for ‘waste’) (Rother and Shook 1998). It helps to effectively design a lean production system. VSM includes creating a map of the flow of material as needed to make a family of products through their production steps, and the flow of information from the customer back to each production process. A current-state map of in-plant value streams then serves as the basis for developing future-state maps that leave out wasted steps while pulling resources through the system and smoothing flow. The difference between the current state and potential future states provides a road map to start the implementation of performance improvements. The scope of many VSMs has been restricted to remain within the boundaries of a single organization. Recent efforts (e.g., Jones and Womack 2002) apply VSM on a macro scale, considering supply-chain upstream and downstream of a specific organization. Adopting such a view is most appropriate in the highly fragmented AEC industry. One value-based metric is the actual work time vs. the total time a products spends in a system, also known as value added time over lead time. An application and analysis of this metric in the SC of pipe supports is presented in sections 7.10.6 and 7.10.7. Batch size: Batch size refers to the unit of handoff of information or work from one SC participant to the next. An analysis of the impact of batch sizing on SC performance is presented in section 7.10.5. Section 7.10.8 then describes different computer-based simulation models based on the SC of pipe supports to illustrate the effects of batching, multitasking, and variability as contributors to lead time. Cost: Besides time, value, and quality, cost is an important metric. However, cost data is more sensitive to obtain and the researchers did not insist on doing so. Other considerations in comparing different SC configurations may include, to name but 13 Updated 5 June 2002 14 of 36 - 3/3/2016 a few: the distance or directness of control and communication between the various SC participants and the number of process steps in the SC (length of the SC). 7.10.5 Analysis of Various SC Lead Times and Batch Sizes To illustrate the use of various lead time metrics, data was analysed on 680 pipe hangers and supports from one power plant currently under construction. In this case, the SC of pipe supports followed the structure of SC configuration 1 (engineering firm designs pipe supports). Figure 7 represents several lead times for different SC activities and handoffs, starting with the date at which the first purchase order was issued and ending with the actual shipment date to the site. For example, the handoff between design and fabrication is represented by the number of calendar days since the engineering firm issued the drawings to the supplier (‘Issued’) until the support supplier sent detail drawings for approval (‘SA’). This handoff took more than 8 weeks on average (61 days = 8.7 weeks), whereas the fabrication process itself took no more than 6 weeks (27 days + 13 days). To the extent it could be determined, this particular handoff took so long because of the interdependence between participants in the SC and the high degree of analysis and verification required after each handoff, in part due to lack of product and process standardization. Issued SA 61d 18d 18d RF - 'Issued' ref ers to the date when the engineering f irm issued the drawings to the supplier. - 'SA' ref ers to the date at which the supplier sent in shop drawings to the engineering f irm. - 'RF' ref ers to the date at which supports were ready f or f abrication. - '1st Ship' ref ers to the shipping date originally promised by the supplier. - 'Schedule Ship'is the scheduled shipping date, a date af ter the f irst ship date if the supplier missed their earlier promise. - 'Act. Ship' ref ers to the date that supports were actually shipped to the site. - 'Site Need Date'is the date obtained f rom construction to support their original plan. 8d 10d 8d 1st Ship 27d 13d 13d Sched. Ship 19d 15d 19d Act. Ship 6d 2d 6d Site Need Date 32d Legend d=calendar day s Mean Dev . Standard Average Pipe Support Delay Figure 7: Lead Times in Pipe Support Detailing, Fabrication, and Delivery 14 Updated 5 June 2002 15 of 36 - 3/3/2016 7.10.6 Analysis of Value-added Time over Total Delivery Lead Time 7.10.6.1 Value Stream Map Using VSM, the flow of work was documented throughout the design, procurement, detailing, and fabrication phases of pipe supports based on SC configuration 1, which is most commonly used by practitioners in the power plant industry today (Figure 2). The presented value stream analysis did not consider the installation of pipe supports on site because we were unable to get real project data to support the case study. Based on configuration 1, we analyzed the value stream for the supply chain as a whole (this Section) and then detailed the fabrication phase (section 7.10.7). The analysis presented here focuses on the evaluation of value added and non-valueadded times. The results are complemented with observations about the execution of the SC activities and the behavior of the SC participants that possibly cause waste along the chain. Figure 8 depicts the value stream each single pipe support follows from design to delivery to the site. Note how different this map is, with its focus on the flow of a single support, as compared to, for instance, a project-based CPM schedule! This VSM shows a series of linked activity boxes with triangles in between. In terms of duration, activity boxes represent the time a pipe support will be in process in a conversion activity. This time gives an upper-bound estimate of value-added time. The triangles represent the time a pipe support waits until it gets processed by the next activity. This waiting time may different causes that will be explained in Section 7.10.6.3 and 7.10.8 in this report. Triangles do not have any specific duration. Instead the VSM shows total durations between activities (arrows at the top of the activities). Accordingly, the difference between the total time in the system (sum of times shown on arrows) and the processing time (sum of times shown under each activity box) represents a lower bound on the total non-value-added time or waste. The value-added times were determined using two sources of information: factual data from a specific power plant project and more anecdotal data obtained by interviewing tens of piping engineers and pipe support designers from engineering firms as well as support suppliers. 15 Updated 5 June 2002 16 of 36 - 3/3/2016 The unit of value-added time is ‘man-hour.’ Since more than one person may contribute to the completion of any one activity, the real time needed to perform an activity may differ from the value-added time shown, in function of the resource allocation. The total time in the system needs to be understood as the time that a pipe support takes to flow through the SC from start to end. The unit of time in the system is a week, considering that each week corresponds to 40 hours of work per person. 8 weeks Route Pipe 2-2.5 m -hrs 2- 3 weeks 2- 3 weeks Loc ate Pi pe Supports Analyze Pi pe Stres s Des ign Pi pe Support Chec k Loads and Interferenc e 0.5 m-hrs 1.8-2.3 m-hrs 0.5-1.0 m-hrs 3.5 m-hrs 2 weeks 1 week Reinforc e Struc ture (inc . New Details ) Prepare Pi pe Support Drawings Selec t Suppli er and Send Info Analyze Info from Eng. Firm Agree on Pric e (Inc . Budget) 3.5- 6 m-hrs 1m -hrs 1- 2 m -hrs 0.4- 0.6 m-hrs 0.1- 0.5 m-hrs 2- 4 weeks 1- 3 weeks 6- 8 weeks Is s ue Pipe Support Details for Fab Approve Drawings Fabric ate 2-5 m-hrs 1-2 m-hrs 24 m-hrs 1- 2 weeks 2 weeks 1 week Supports Ready to Ship Deliver Supports On Site T otal Duration = 28 -37 weeks Hours per week = 40 hrs . T otal T ime in Sys tem = 1120 - 1480 hrs 1 m -hrs Duration (hrs) Notes: 1. 2. - All durations are per uni t of s upport (m-hrs = man hours ). Underlined values have been as s um ed by res earc hers . T he num ber of m an-hours for the ac tivity "Agree on P ric e" was as s um ed as 0.1-0.5 m -hrs /s upport. T he num ber of m an-hours for the ac tivity "Fabric ate" was as s umed as 24 m-hrs /s upport. T he num ber of m an-hours for the ac tivity "Del iver" was as s um ed as 1 m-hrs /s upport. In all c as es , the queue times are s o big than thes e as s umpti ons won't affec t the final res ults . The Supplier performs the Activity T otal Queue T ime = 1078-1428 T otal P roc es s ing T ime = T otal T ime in Sys tem 42-52 = 1120-1480 % 96 - 96.5% 4 - 3.5% 100% Interaction between Engineering Firm and Supplier to perform the Activity The Engineering Firm performs the Activity Figure 8: Value Stream Map - Supply Chain of Pipe Supports 7.10.6.2 Value-stream Analysis Results Analysis of this VSM shows that a pipe support takes a total duration ranging from 28 to 37 weeks to flow through the system. One reason for this variation is the diversity and complexity of supports that are covered by the design, detailing, and fabrication phases as shown in Figure 8. The analysis shows that only about 4% of the total time a pipe support needs to flow through the system represents value-added time. This means that only about 1.6 hours out of a 40-hour work week really add value to the final pipe support product. The remaining 96% of the time or 38.4 hours out of a 40-hour work week 16 Updated 5 June 2002 17 of 36 - 3/3/2016 represent non-value-added time. 7.10.6.3 Causes of Non-value-added Time Causes of waste in this particular supply chain are mainly related to the time resources (information and materials) wait to be processed and the amount of rework in the system. Wait time in part stems from batching practices. Batching is an important consideration in supply chain performance assessment because bigger batch sizes cause longer waiting times and therefore longer lead times. This case study identified several types of batches with different sizes along the supply chain. Figure 9 depicts two of these, namely (1) the release of design information from the engineering firm to the support supplier, and (2) the shipment of completed supports from the support supplier to the site. As shown, some batch sizes on this project were as big as 260 supports. This means that the first support had to ‘wait’ for the other 259 supports to be processed, until all were released to the next activity. 280 260 240 Batch Size Delivery of Supports to the Site Release Design Info to Supplier 220 200 180 160 140 120 100 80 60 40 Days 20 0 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 Figure 9: Example of Batch Size Variation Other contributors to wait time are multitasking practices. In reality, piping engineers and designers, often aided but also impeded in some way by the tools they need to perform design activities (e.g., programs require particular sets of input data and won’t run with data missing), multitask between two or more design processes that may belong to one or more power plant projects. They do so because not all information or other resources 17 Updated 5 June 2002 18 of 36 - 3/3/2016 needed to complete a task may be available when they have the time to work on that task. Multitasking enables them to reduce their own idleness, though it does not necessarily increase their efficiency because each switch of tasks comes at a setup cost. Unfortunately, more multitasking means that any one task has a smaller likelihood of being worked on and completed and, consequently, it has to wait longer and its lead time increases. Management controls multitasking practices by setting execution priorities. Tasks with a higher execution priority are ‘expedited’ at the expense of those with a lower execution priority, which then have to wait longer and get stretched in lead time. Finally, rework is due to uncertain data being incorporated in the early design phase of pipe supports or other supply chain phases. Rework may be due to several factors but especially rework due to errors should not be tolerated as it reduces throughput, the time to make (design, procure, fabricate, and deliver) a pipe support, and it causes unreliable workflow. A major rework factor is working from vendors preliminary information, only to find that connection locations, sizing and ____________have changed. 7.10.7 Detailed Value Stream Map for Fabrication In Learning to See, Rother and Shook (1996) provide a step-by-step process to analyze not only the production flow but they also emphasize lean tools to support the workflow along the different workstations. We have mapped the fabrication of engineered pipe supports but still lack information to provide a detailed map like theirs for any one specific product family. Instead, the fabrication process as studied here focuses on the analysis of value-added and non-value-added times. Figure 10 depicts a VSM for the fabrication of pipe supports. The fabrication lead time quoted to us by practitioners varies from 2 to 6 weeks. One reason for this variation is, again, the diversity and complexity of supports. The number of man-hours per week, considering a three-shift operation, is 7 days/week * 24 hours/day, which equals 168. The total number of hours that a support remains in the system until it is delivered to the site thus varies from 336 to 1,008. In Figure 10, value-added times are shown under each activity box: the total value-added time is equal to 107.38 hours. This represents between 11% and 32% of the total time that a pipe support remains in the system and it means that 18 Updated 5 June 2002 19 of 36 - 3/3/2016 only 1-to-3 out of 10 hours of work in the fabrication shop really add value to the final pipe support product. Cut Roll/Bend Drill Fitup Weld 2.5 m-hrs 0.33 m-hrs 1 m-hr 40 m-hrs 40 m-hrs Start Fab Total Duration = 2 - 6 weeks Hours per week = 168 hrs. Total Time in System = 336 - 1008 hrs Durations 1 (hrs) % 231 - 901 69 - 89% Total Processing Time = 105 - 107 31 - 11% Total Time in System 336 -1008 Total Queue Time = = 100% Pack Final Inspection Paint or Galvanize 2-4 m-hrs 0.5 m-hrs 1 m-hr The Supplier performs the Activity Specialty Contractor may perform the activity. Assemble and Paint 18 m-hrs 2 Notes: 1. All durations are per unit of support (m-hrs = man hours). 2. Some suppliers use to paint and galvanize the support as part of their standards but in most cases painting is commonly performed in-house. The galvanizing process is more complicated and usually pipe supports need to be transported to another facility to be galvanized. Specialty contractors perform galvanizing. Figure 10: Value Stream Map - Fabrication Phase of Pipe Supports Note that the durations shown in Figure 10 do not exactly match those shown in Figure 8. The reason is that data was obtained through several interviews with practitioners and each interviewee has developed a mental model of the SC based on their own experience. 7.10.8 Process Simulation Showing Impact of Contributors to Lead Time 7.10.8.1 Value of Using Simulation Computer-based process simulation is used next to highlight several contributors to long lead times, relative to the modest amount of value-added time that is needed to execute each value-adding activity in the SC for the delivery of capital projects. Contributors to long lead times are worthy of study because they indicate opportunities for process improvement. The contributors discussed here tend not to be recognized in traditional project management practice. Traditional project management relies on mostly deterministic- rather than stochastic models (such as models that acknowledge an activity’s duration may not be precisely know, but instead could fall within a range of values) that further extend in scope to include handoffs and flows, and the delivery of value. Simulation, then, helps project managers to quantify waste in current practices and assess opportunities for schedule compression. 19 Updated 5 June 2002 20 of 36 - 3/3/2016 Simulation illustrates dynamically how uncertainty, multitasking, and batching throughout various SC phases hamper overall SC performance. Uncertainty is introduced to mimic variability in the system. Multitasking means that a resource is to perform more than one task and (often randomly) alternates working on one or the other(s). Batching means that activity outputs are released to the next activity in groups that follow certain grouping criteria. Individually and in combination, uncertainty, multitasking, and batching all cause an increase in SC lead time. 7.10.8.2 Simulation of Design Process for Pipe Supports The aim of the simulation models that are described next is to illustrate how interfaces between SC tasks affect overall system performance, though not to detail any specific value-adding task(s) in the SC. The models accordingly cover only an excerpt (Figure 11) of the process model previously shown in Figure 8. The model could easily be extended to span the entire process presented in Figure 8, but this was not done in order to keep the discussion of modeling outputs straightforward. The excerpt focuses on routing pipe, locating pipe supports, and conducting a pipe stress analysis. It reflects data collected on several power plant projects. Each of the tasks, respectively, takes about 2-to-2.5 manhours (mh), 0.5 mh, and 1.8-to-2.3 mh, or a total of 4.3-to-5.3 mh of value-added time for each support to be designed. However, practitioners say that a lead time on the order of 8 weeks for a support to go through this sequence of design steps is required. The presented simulation models provide a rationale for the causes of this discrepancy, which are essentially related to task variability, multitasking, and batching. Other factors beyond these may also contribute to this discrepancy though they are not included in the discussion. 8 weeks Route Pipe Locate Pipe Supports Analyze Pipe Stress 2-2.5 m-hrs 0.5 m-hrs 1.8-2.3 m-hrs Figure 11: Excerpt of Value Stream Map for Design of Pipe and Pipe Supports 20 Updated 5 June 2002 7.10.8.3 21 of 36 - 3/3/2016 Design Process Model in STROBOSCOPE The simulation models in this report build on the STROBOSCOPE discrete-event simulation engine (Martinez 1996) and illustrate several scenarios that combine different uncertainty levels, different degrees of resource sharing, as well as different batch sizes. Table 1 summarizes the functionality of the symbols used in these models. All models were run using STROBOSCOPE (version 1,2,2,0) on a Pentium 566-Mhz. Computer running Windows 98 second edition. Table 1: Selected STROBOSCOPE Symbols SYMBOL NAME Queue Normal (activity) Combi (-nation activity) EXPLANATION Is a holding place (buffer) for 0, 1, or several resources waiting to become involved in the succeeding combination activity. Queues may contain generic or characterized resources. The latter are distinct from one another and they can be traced as individuals through various network nodes during simulation. The logic describing the ordering of resources upon entry into a queue of characterized resources is termed a DISCIPLINE. Describes a certain type of work to be done, or a delay, of a known (probabilistic) duration from start to finish. May require a single resource or no resource at all. Like a normal, describes a certain type of work to be done, or a delay, of a known (probabilistic) duration from start to finish. Unlike a normal, requires several resources in combination for its performance and draws what is needed from the queue(s) that precede it. Consolidator Acts as a counter up to n (n is an integer value specified with the node): after n resources have been released into the consolidator, all n resources at once will be released from it. Link Shows flow logic. Should be labeled to meaningfully describe the resources that flow through it. If the link emanates from a queue, a DRAWORDER may be specified to sequence resources being drawn from the queue. Figure 12 depicts the STROBOSCOPE process model that corresponds to the value stream map shown in Figure 8. The most relevant modeling assumptions are the following: a. The process model includes (1) a primary chain with RoutePipe1, Locate1, and AnalyzeStress1 (surrounded by a blue solid box in Figure 12) and (2) a secondary chain with RoutePipe2, Locate2 and AnalyzeStress2 (surrounded by a red dashed box). For a specific project being studied, it is assumed that pipe supports will 21 Updated 5 June 2002 22 of 36 - 3/3/2016 flow through the primary chain. The purpose of the secondary chain is to illustrate that resources engaged in the primary chain also perform other work for other projects. b. The inputs and outputs of each task in the secondary chain are independent of those of other tasks: they are decoupled (e.g., the output of RoutePipe2 does not yield direct input into Locate2, whereas that of RoutePipe1 fed into Locate1) to reflect that the amount of work designers have on different projects can vary substantially. c. The effect of batching is introduced only in the primary chain since the secondary chain has been decoupled. This effect is included in the design and fabrication phases using three different consolidator nodes called BatchRoute, BatchLocate and BatchStress. d. The duration of each task may vary over a range of values, which mimics variability in the design of various kinds of pipe supports. RoutePipe2 GENERATE 1 RouteTwo PIPING ROUTING AResource2 INIT 1 P4 ReadyLo cate2 ARoute2 Info3 ATextFile3 INIT 100000 P5 F11 BatchLocate BS=Variab le Resourc e2 LINK LABELS Flow of Information 1 Flow of Information 2 Stress1 s1 ALocate1 AResource3 INIT 1 Locate2 P8 GENERATE 1 LocateTwo Readyto Stress2 Info4 ALocate2 ATextFile4 INIT 100000 PIPE SUPPORT LOCATION Primary Chain F P F12 Readyto F13 AnalyzeStres F16 CHARACTERIZED RESOURCES TYPE SUBTYPE ATextFile1 InfoOne ATextFile2 InfoTwo ATextFile3 InfoThree ATextFile4 InfoFour ARoute1 RouteOne ARoute2 RouteTwo ALocate1 LocateOne ALocate2 LocateTwo BatchStress BS=Variab le F17 Done1 AStress1 Resourc e3 P10 ATextFile2 INIT 100000 ARoute1 P6 P1 BS = Batch Size Locate1 cate1 BS=Variab le Resourc e1 P2 Info2 F7 ReadyLo F8 F14 AResource1 INIT 1 BatchRoute F9 ATextFile1 F6 GENERATE 1 StressOne F15 RoutePipe1 Consolidate LocateOnes P11 F3 Info1 F4 AStart INIT 1 F2 GENERATE 1 LocateOne F10 Priority F5 F1 Consolidate RouteOnes P3 Start 1 GENERATE 1 RouteOne P8 GENERATE 40 InfoOne P9 AnalyzeStres P12 Done2 s2 GENERATE 1 StressTwo AStress2 PIPE STRESS ANALYSIS Secondary Chain CHARACTERIZED RESOURCES TYPE SUBTYPE AResource1 ResourceOne AResource2 ResourceTwo AResource3 ResourceThree AStress1 StressOne AStress2 StressTwo Figure 12: STROBOSCOPE Process Model of for Pipe and Pipe Support Design Tasks 22 Updated 5 June 2002 7.10.8.4 23 of 36 - 3/3/2016 Implementation and Simulation of Supply Chain Tasks Using Figure 12’s graphical representation of supply chain tasks, deterministic and probabilistic simulation scenarios were implemented. The resources that remain constant in these scenarios are: a. The number of pipe supports to be designed for the project flowing through the primary supply chain (represented by Info1) is equal to 40 units. This number corresponds in order of magnitude to the number of supports that are engineered to suit the main steam system of a power plant project. It is sufficiently large to yield interesting simulation results yet sufficiently small for the simulation processing time to remain small. b. The number of pipe supports that enter the secondary supply chain is equal to 100,000 units. This number is set to be very high to reflect the assumption that the design firm has a lot of work to do. We ignore the project schedules that define due dates on any of their design tasks; otherwise, they would affect the prioritisation of work. c. The number of resources Resource1, Resource2, and Resource3 is set equal to 1 unit for all models. These resources will be shared by the primary and the secondary chain in scenario 2. Table 2 summarizes the two scenarios that are detailed next. Table 2: Simulation Scenarios Scenario Type of Model Chain Focus 1 Deterministic Primary Batching 2 Probabilistic Primary + Secondary Batching/Variability/Multitasking 7.10.8.5 Scenario 1: Deterministic Model with Batching The first simulation scenario illustrates the contribution of batching to lead time. Only the primary chain with deterministic durations is considered in this model. This model serves as a baseline for comparing behaviour against other, probabilistic simulation scenarios that combine uncertainty, multitasking, and batching. 23 Updated 5 June 2002 24 of 36 - 3/3/2016 Scenario 1 has been simulated considering different batch sizes as listed in Table 3. In all cases, the durations of the tasks RoutePipe1, Locate1, and AnalyzeStress1 are 0.28 (2.25 mh / 8 mh = 0.2812), 0.06 (0.5 mh / 8 mh = 0.0625), and 0.26 (2.05 mh / 8 mh = 0.2562) working days per support respectively. These values represent the data shown in Figures 8 and 11, converted into the equivalent number of working days per support (Table 4). Table 3: Model Parameters and Outputs for Scenarios 1 and 2 Simul. Run BatchRoute [number of supports] BatchLocate [number of supports] BatchStress [number of supports] Scenario 1 Duration [work day] 1 2 3 4 5 6 7 1 2 4 8 10 20 40 1 2 4 8 10 20 40 1 2 4 8 10 20 40 11.52 11.84 12.48 13.76 14.40 17.60 24.00 Scenario 2 Duration based on 100 simul. runs [work day] Mean Standard Dev. 11.69 11.88 12.54 13.72 14.45 17.52 24.01 0.32 0.29 0.30 0.31 0.30 0.23 0.34 Table 4: Activity Durations in Models for Scenarios 1 and 2 Task Scenario 1 Task Durations [work days / support] RoutePipe1 RoutePipe2 Locate1 Locate2 AnalyzeStress1 AnalyzeStress2 0.28 n/a 0.06 n/a 0.26 n/a Scenario 2 Task Durations [work days / support] Normal [mean, standard deviation] Normal [0.28,0.05] Normal [0.28,0.05] Normal [0.06,0.01] Normal [0.06.0.01] Normal [0.26,0.05] Normal [0.26,0.05] The results of these simulation runs for scenario 1 are plotted in Figure 13. The relation between batch sizes (which, for convenience were chosen to be the same for each of the three tasks) and lead time here is linear. As is to be expected, the worst situation arises when the batch size is equal to the total number of supports that enter into the system. This situation results in the longest lead time. In this case, a batch size of 40 at each task results in a lead time more than twice the lead time for a batch size of 1. This plot demonstrates that batching is an important consideration in SC design because the bigger the batch size, the longer the lead time of the process overall. While 24 Updated 5 June 2002 25 of 36 - 3/3/2016 this finding is nothing new in the field of production management (e.g., Hopp and Spearman 2000), design and project managers of construction project are not necessarily aware of it, or, accordingly, consciously shaping batch sizes with SC performance in mind. 40 B a t c h S iz e ( P ie c e s ) 35 30 25 20 15 10 5 L e a d T im e ( W o r k in g D a ys ) 0 0 5 10 15 20 25 30 35 40 45 50 55 60 Figure 13: Effect of Batch Size on SC Lead Time 7.10.8.6 Scenario 2: Probabilistic Model with Batching, Variability, and Multitasking Table 4 presents the probabilistic task durations that are used to simulate variability in the system for scenario 2. For example, based on probability, a normal distribution applied to RoutePipe1means that 68% of the time, the task duration will fall within the range delimited by the distribution’s mean +/- standard deviation. Specifically, 68% of the time, a duration will be greater than 0.23 (0.28-0.05) working days/support but less than 0.33 (0.28+0.05) working days/support. Scenario 2 illustrates the impact of batching combined with variability and multitasking on lead time. The secondary chain now has been added to the model in order to mimic the effect of multitasking in the system. Table 3 presents the outputs from this model corresponding to the batch-size combinations described in the same table and the extreme case of 100% task priority (which, in effect, means no multitasking). This output was computed using data from 100 random simulation runs. 25 Updated 5 June 2002 26 of 36 - 3/3/2016 In reality, engineers multitask between two or more design processes. Multitasking, then, needs to be controlled by execution priorities. In this scenario, the simulation engine is programmed to share resources based on pre-determined priorities. Several simulations have been run considering priorities ranging from 10% to 100%. Based on Figure 12, priorities are assigned to tasks in the primary chain so that a priority value of 30% reflects that when a resource becomes available, it will randomly select to work on that task 30% of the time, and on competing tasks (tasks in the secondary chain) the remainder 70%. Figure 14 illustrates that lead times generally increase with an increase in batch size and with a decrease in priority of the primary chain. For example, a batch size of 30 supports and a priority of 80% for the primary chain (P = 0.8), results in an average lead time of about 23 working days. A batch size of 30 but a priority of 50%, results in an average lead time of 54 working days (more than double the previous lead time). Figure 14: Lead Time (mean value ± standard deviation) vs. Batch Size for Different Multitasking Priorities (P) of Primary SC, Using Data from 100 Random Simulation Runs Figure 15 shows the effect of multitasking from a different perspective. Clearly, lower priorities yield increasingly longer lead times. These models describe how a system for the delivery of capital projects can be viewed as a series of tasks that depend on other tasks for handoffs occurring at discrete times. In order to illustrate how handoffs affect SC performance, Figures 16 through 18 26 Updated 5 June 2002 27 of 36 - 3/3/2016 graph the outputs of three different simulation runs showing the timing of handoffs between tasks in the primary chain, based on different batch sizes and a 50% multitasking priority or the primary over the secondary chain. Clearly, batching and multitasking affect SC lead time. Figure 15: Lead Time (mean value ± standard deviation) vs. Multitasking Priority of Primary SC for Different Batch Sizes (B) In terms of lead time, the ideal situation is created when the batch size for each task is 1, so that the handoffs are frequent, the flow is smooth, and the overall SC incurs the least delay. This ideal situation is not practical, though, because of setup times that make it more rational for the ‘optimal’ batch (economic lot size) in any one process to be greater than 1 unit. Figure 16 illustrates simulation output when all task handoffs occur in unit batches of 1. It shows a lead time of 35 work days or 7 weeks. This is the shortest lead time obtainable with the current configuration, but notice that the system is unbalanced: tasks progress each at a different pace. The actual pace is indicated by the slope of each line in Figure 16. The maximum possible pace is indicated by the duration of the task as given in Table 4, for instance, Locate is the fastest task of all. Because RoutePipe is slower but hands off an output to it, Locate can only go as fast as RoutePipe in this configuration. 27 Updated 5 June 2002 28 of 36 - 3/3/2016 Figure 16: Lead Time vs. Number of Supports for Batches of 1 Support Figures 17 and 18 illustrate the effect of batch size on performance in the primary supply chain. Figure 17 considers three different batch sizes: respectively 10, 10, and 20 supports for the activities RoutePipe1, Locate1, and AnalyzeStress1 (these numbers were chosen arbitrarily). Each vertical line in the chart represents an output batch being handed off as input to the next task. For instance, figure 17 shows that Locate1 outputs units that accumulate into a batch of 10 units relatively fast. Then, it has to wait until RoutePipe1 releases more output before it can work on the next batch handed to it. During this wait, Locate1 multitasks with Locate2 in the secondary chain or simply remains on stand by. Figure 17 shows that the larger batch sizes result in a larger lead time, now reaching 42 work days. Figure 18 considers the largest possible batch size for this model, in this case 40 supports, resulting in a lead time of 59 work days, significantly greater than the 35-day lead time obtained with unit handoffs. 28 Updated 5 June 2002 29 of 36 - 3/3/2016 Figure 17: Lead Time vs. Number of Supports for Varying Batch Sizes (10 for RoutePipe, 10 for Locate, and 20 for AnalyzeStress) Figure 18: Lead Time vs. Number of Supports for Batches of 40 Supports 7.10.8.7 Summary of Simulation Findings Computer-based simulation models were included in this case study in order to illustrate how such modelling efforts can help in conducting what-if analyses to assess the performance of alternative SCs, based on different configurations and management 29 Updated 5 June 2002 30 of 36 - 3/3/2016 decision making (e.g., resource allocation and assignment of task priorities, reduction of variability, and sizing of batch). The simulation models presented here were limited in scope but could easily be extended to capture any additional complexity. Note that prioritising tasks and, accordingly, allocating resources to multitask so as to make best use of them is—in and by itself—not necessarily a bad practice. However, scenario 2 illustrated that the randomness with which a person may switch from one task to another is what interjects uncertainty into the system, and that uncertainty harms overall throughput performance. In addition, the model does not include any penalty for switching from one task to another one, though in reality there always will be some setup or cleanup time associated with each switch. In any case, comparison of simulation outputs can help to assess what the likely values would be of metrics when a specific SC actually gets implemented. The VSM shown in Figure 11 showed the direct work for each task per pipe support and also the 40 work-day (8 weeks * 5 work days/week) total lead time to get a support through the system. This direct work defined the task durations in the simulation models. Given the direct work alone, and assuming no imbalances in task rates or batching or the like, one might compute the lower bound on the total duration to get 40 supports through the system: it is equal to 40*2.5 mh + 0.5 mh + 2.3 mh = 102.8 mh or 102.8 mh / 8 mh/day = 12.85 work days. The discrepancy between these 13 work days and the 40-day lead time may be explained by a scenario like that depicted in Figure 17 (lead time of 41 days) that shows the effects of batching, variability, and multitasking on total lead time. Of course, these explanations must be treated with caution, as they are not the only ones plausible. 7.10.9 Supply Chain Improvements 7.10.9.1 SC Improvements Implemented to Date SC improvement can be achieved by implementing any or all SC tactics as described in Chapter 5 of this Research Report. The company that initiated this case study has implemented several, namely: 1. Select a supplier for long-term collaboration on multiple projects and involve that supplier early in design. The company in this case investigated alternative 30 Updated 5 June 2002 31 of 36 - 3/3/2016 suppliers of supports for the kinds of capital facilities they consider to be part of their core competence, then selected one and worked out a longer-term agreement with them, which included (2) and (3). This agreement then eliminates a seven week duration period normally required to issue an RFP to several pipe support firms, receive and analyze bids, negotiate and award a P.O. 2. Standardize the design of pipe supports (standardize product design). 3. Use electronic data interchange and standardize engineering-supplier interface processes (standardize process design). Because hard data was not available to quantify performance metrics and document actual SC improvement, figures 19 and 20 show speculative data on the improvements that might be expected. Figure 19 presents a shortened version of the VSM depicted in Figure 8. By selecting a supplier early, not only can most of the direct work (1-2 mh/support) associated with this task ‘Select Supplier and Send Info’ be taken out of the SC, more importantly, the lead time associated with it (1 week) also gets taken out (see the cross-out task in Figure 19). Thanks to product and process standardization, the engineering firm as well as the supplier now spend less time on ‘Prepare Pipe Support Drawings,’ “Analyze Engrg. & Prepare Budget,’ ‘Issue Pipe Support Details for Fab.,’ and “Approve Drawings’ (see the circled tasks in Figure 19). It is also likely that ‘Design Pipe Support’ and ‘Fabricate’ are favorably impacted by standardization (see the dashedcircled tasks in Figure 19). Again, these changes not only remove direct work from the SC, they also shorten the lead times corresponding to each task, and the latter has an even greater impact than the former on overall reduction in SC lead time. After implementation of these selected SC improvement initiatives, the total lead time of 28-37 weeks, gets reduced by an estimated 20% to a range of 23-30 weeks. The total amount of direct work has decreased by an estimated 7.5% and is down by 2.5-4.2 mh, so that the ratio of value-added work/total lead time increases from 3.5-4% (Figures 8 and 19) to 4-4.3%% (Figure 20). Note that this ratio is still small and therefore suggests that numerous other improvements yet remain to be implemented. Nevertheless, the gains in lead time improvement, obtained thanks to the implemented SC tactics, are impressive. 31 Updated 5 June 2002 32 of 36 - 3/3/2016 8 weeks 2-3 weeks 2-3 weeks 2 weeks 1 week Design Piping System Design Pipe Support Check & Modify Other Systems Prepare Pipe Support Drawings Select Supplier and Send Info 4.3-5.3 mh 0.5-1.0 mh 7-9.5 mh 1 mh 1-2 mh Start 3 weeks 2 weeks 2 weeks 6-8 weeks Analyze Engrg. & Prepare Budget Issue Support Details for Fab Approve Drawings Fabricate 1-2 mh 24 mh 0.5-1.1 mh 2-5 mh 1 week Supports Ready to Ship Deliver Supports On Site 1 mh Total Lead Time = 29-33 weeks or 1160-1320 hrs (at 40 hrs/week) Total Value Added Time = 42.3-51.9 hrs VAT/Lead Time = 3.5-4% Figure 19: Tasks Impacted by SC Performance Improvement Initiatives (Early Supplier Selection and Involvement in Design, Product Standardization, and Process Standardization) Start 8 weeks 2-3 weeks 2-3 weeks Design Piping System Design Pipe Support Check & Modify Other Systems 4.3-5.3 mh 0.5-1.0 mh 7-9.5 mh 4-7 weeks Engineer and Supplier Collaborate 3-8 mh 6-8 weeks Fabricate 24 mh 1 week Supports Ready to Ship Deliver Supports On Site 1 mh Total Lead Time = 23-30 weeks or 920-1200 hrs (at 40 hrs/week) Total Value Added Time = 39.8-47.8 hrs VAT/Lead Time = 4-4.3% Figure 20: SC after Implementation of Selected SC Improvement Initiatives 7.10.9.2 On-going Improvement Efforts To further improve SC performance, the company in this case study is now increasingly using modularization and offsite assembly of entire structural steel bridges with complete piping systems. These practices reflect a trend in the industry (e.g., Burke and Miller 1998, Schimmoller 1998, Gotlieb et al. 2001). Burke and Miller (1998) depict two examples. Figure 21 illustrates a “modularized pipe rack being moved into place at the 32 Updated 5 June 2002 33 of 36 - 3/3/2016 Mid-Georgia Cogeneration Plant. Piping, hangers, instrumentation, and the auxiliary boiler deaerator are already integrated into the pipe rack modules.” Figure 22 illustrates that “modularized pipe rack systems were incorporated into the design and construction of the Cleburne Cogeneration Facility” Figure 21: Modularized Pipe Rack at MidGeorgia Cogeneration Plan (photo courtesy of Black&Veatch, reprinted in Burke and Miller 1998) Figure 22: Modularized Pipe Rack at Cleburne Cogeneration Plan (photo courtesy of Black&Veatch, reprinted in Burke and Miller 1998) The case-study company’s efforts are substantially motivated by a contract to consecutively build five power plants of the same cookie-cutter design for one owner. The use of reference power plants has been common in the EPC industry since the 1970s, but the EPC firm’s ability to tie these products to SC process execution, including longerterm agreements with various suppliers, has traditionally been hampered by their inability to forecast demand. Owners have a significant role to play in driving further SC improvements of this kind, by providing EPC firms with transparency into their demand forecasts and contracting for multi-project procurement SMC standardization will result in a lower TIC. 33 Updated 5 June 2002 7.10.10 Conclusions 7.10.10.1 Case Study Conclusions 34 of 36 - 3/3/2016 This case study has illustrated that construction SCs are intrinsically complex and varied by presenting alternative SC configurations that may be used in the delivery of pipe supports used in power plants. Power plant owners and/or engineering firms may independently or jointly select any one or several configurations to suit a specific project’s requirements. Suppliers can significantly influence the shape of these SC by offering a judiciously selected range of standardized products, showing willingness to collaborate with designers on solution development, and possibly providing integrated SC services for design, pipe support (and other plant component) fabrication, as well as construction. A benefit of having several SC configurations to choose from is that the abilities and constraints of SC participants can be balanced in order to achieve project goals. Owners generally leave it up to the EPC firm and the supplier to manage the selected configuration(s). The presented case study has demonstrated that managing the interfaces between SC tasks can yield significant performance improvements, not only by reducing the direct work performed by SC participants, but more dramatically, by improving the way handoffs occur from one task to the next. The application of value-stream mapping, computer-based simulation, and metrics have shown to be useful tools to start SC performance improvement initiatives. Gains of 20% reduction in SC lead time and 7.5% reduction in direct work appear feasible thanks to the implementation of SC tactics such as early supplier selection and their involvement in design, combined with product and process standardization. The value-added time/lead time metric, found to have a value of about 4% in this case study, suggests that there is significant opportunity for further improvement. 7.10.10.2 Improvement Areas worth Further Investigation A number of other SC tactics should be considered to achieve further improvement in performance of the SC for pipe supports but a few stand out as worthy of getting priority for further investigation. 34 Updated 5 June 2002 35 of 36 - 3/3/2016 1. This case has focused on the SC of pipe supports, starting with piping system design. The study has not sought out linkages and issues of synchronization between parallel SCs such as those involving procurement and fabrication of pipe, structural steel, or instrumentation, valves, and fittings; nor those pertaining to delivery or matching of supports with other components needed in combination to allow for installation at the site (e.g., Tommelein 1998). Nevertheless, construction is a key beneficiary of excellence in SC execution. Data collection regarding means to help synchronize SCs in light of design and schedule changes, and further study of alternative SC configurations are in order to increase understanding of these performance issues and recommend improvements. 2. Additional research is needed to relate modularization efforts to improved SC performance. 3. A study is needed of the effect of ‘commoditization’ of engineering services as a contributor to SC performance improvement (also see ENR 2002). 4. The emergence of the role of supply chain integrator and the value people in this role bring to SCM must be recognized. This role must be defined in the context of functions currently filled at the project- and corporate level within organizations and at the supply-chain level across organizations. References 35